On 27 July 2022, a three-judge bench of the Supreme Court of India delivered the judgment in Vijay Madanlal Choudhary v Union of India and, in a single ruling of more than five hundred pages, upheld almost every contested provision of the Prevention of Money Laundering Act, 2002. The powers of the Directorate of Enforcement to attach property before trial, to search premises and persons, to arrest without a prior judicial warrant, to summon any person and record a statement usable in evidence, to hold an accused to bail conditions that reverse the ordinary presumption of innocence, and to place the burden of disproving the character of attached property on the accused were all sustained. Then, on 25 August 2022, twenty-nine days later, a bench presided over by the Chief Justice of India issued notice on review petitions against that very judgment and recorded that at least two of its conclusions appeared to require reconsideration. A judgment that upheld a statute across the board was therefore flagged by the institution that delivered it, within a month, as partly unsettled. Understanding the Vijay Madanlal Choudhary judgment means holding both of those facts at once, and it also means understanding the argument about conviction rates that surrounds the statute, an argument in which both sides quote true numbers that appear to contradict each other and do not.

The interpretive question at the centre of the case was not whether money laundering should be an offence. No petitioner argued that. The question was whether a statute built to trace concealed money could lawfully be operated through machinery that departs from the ordinary criminal process at almost every point: investigation by officers who are not police officers, an initiating document the accused has no right to see, statements recorded under compulsion that remain admissible, property frozen on an officer’s recorded belief before any court has heard the matter, arrest on the same belief, bail available only when the court is satisfied the accused is not guilty, and a trial at which the accused carries the burden of proving that property is untainted. The statutory text created each of those departures distinctly enough; what it did not settle was whether the Constitution permits them, singly and in combination. That is the question the bench of Justices A. M. Khanwilkar, Dinesh Maheshwari and C. T. Ravikumar, in a judgment authored by Justice Khanwilkar, had to answer, and it answered it, provision by provision, largely in the statute’s favour.

The provisions the Court was asked to strike down

The Prevention of Money Laundering Act, 2002 was enacted as Act 15 of 2003 and brought into force on 1 July 2005. Its design, and the detail of its sections, belong to the statute’s own pillar article in this series, the complete guide to the PMLA, which owns the provision-by-provision account. What this article must fix is the set of provisions whose constitutional validity was in issue, because the holdings cannot be understood without knowing precisely what was attacked.

The attack ran across the architecture of the Act. Section 3, which defines the offence of money laundering in terms of any process or activity connected with the proceeds of crime, including concealment, possession, acquisition, use and projecting or claiming the property as untainted, was challenged for breadth. Section 5, which empowers an authorised officer to provisionally attach property on a recorded reason to believe that it is proceeds of crime and likely to be concealed or transferred, was challenged as executive punishment without trial. Section 8(4), under which the Enforcement Directorate may take possession of attached property once the Adjudicating Authority confirms the attachment, was challenged on the same ground carried a stage further. Sections 17 and 18, governing search and seizure of property and search of persons, and Section 19, governing arrest, were challenged for want of the safeguards that surround the corresponding police powers. Section 24, which obliges a court to presume that proceeds of crime are involved in money laundering unless the contrary is proved, was challenged as a reversal of the presumption of innocence. Section 44, which routes trials to Special Courts, Section 45, which makes offences under the Act cognizable and non-bailable and imposes the two additional conditions on bail, and Section 50, which empowers officers to summon persons and record statements on oath, completed the list. Standing behind all of them was a document the Act does not mention at all: the Enforcement Case Information Report, the internal record by which the Directorate opens a case, whose non-supply to the accused became one of the two questions the Court later flagged for reconsideration.

What exactly was challenged in Vijay Madanlal Choudhary?

The challenge attacked the Act’s operating machinery rather than the offence itself: provisional attachment (Section 5), possession (Section 8(4)), search and seizure (Sections 17 and 18), arrest (Section 19), the reverse burden (Section 24), the twin bail conditions (Section 45), the summons power (Section 50), and the withholding of the Enforcement Case Information Report.

Read that list as a sequence and its logic appears. The petitioners’ case was that the Act does not merely add an offence to the criminal law; it constructs a parallel process in which the ordinary sequence of criminal justice, complaint, investigation by police, charge, trial at which the prosecution proves guilt, is replaced at every stage by something less protective. The summons stage compels answers that an ordinary suspect could refuse. The attachment stage takes property before charge. The arrest stage proceeds on an officer’s belief rather than a magistrate’s order. The bail stage requires the accused to demonstrate innocence. The trial stage presumes the incriminating fact. Each departure, the petitioners accepted, might be defended in isolation; their argument was that the combination produces a regime in which the process itself is the punishment, and in which a person can lose property, liberty and reputation for years without any court having found that an offence occurred.

The Union’s answer, advanced by the Solicitor General, Tushar Mehta, and the Attorney General, K. K. Venugopal, was that money laundering is unlike ordinary crime in a way that justifies each departure. Laundered money moves in seconds, across accounts, entities and borders, and is concealed precisely so that the trail cannot be reconstructed by slow process. A regime that waits for a completed predicate trial before touching property guarantees that the property will be gone. A regime that treats the financial investigator as a police officer, with the bar on using statements made to police, guarantees that the one person who knows where the money went cannot be asked. The statute, the Union argued, was Parliament’s response to obligations India had assumed under international conventions against drug trafficking and transnational organised crime and to the standards of the Financial Action Task Force, and its severity fell on the property and its trail rather than on the person.

The twin conditions: struck down, rewritten, and challenged again

One provision in the batch had a history that the rest did not, and that history explains both the 2022 holding and much of the criticism of it. Section 45, as originally framed and as it stood when the Act commenced, made the twin conditions applicable to a person accused of an offence punishable with imprisonment of more than three years under Part A of the Schedule to the Act. The conditions themselves are severe. Before releasing an accused on bail, the court must give the public prosecutor an opportunity to oppose the application, and where the application is opposed the court must be satisfied that there are reasonable grounds for believing that the accused is not guilty of the offence and is not likely to commit any offence while on bail. A bail court must therefore record, in substance, a preliminary finding of innocence, which inverts the ordinary inquiry into flight risk, tampering and the prima facie strength of the case.

In Nikesh Tarachand Shah v Union of India, reported at (2018) 11 SCC 1, a bench of Justices Rohinton Fali Nariman and Sanjay Kishan Kaul struck those conditions down by a judgment delivered on 23 November 2017. The reasoning matters more than the result. The bench held the provision manifestly arbitrary and violative of Articles 14 and 21 of the Constitution, and the core of the vice, in its analysis, was the classification the provision used. Whether an accused faced the near-insuperable bail bar depended not on anything about the laundering alleged against him but on the punishment attached to the predicate offence listed in Part A of the Schedule. Two persons accused of identical laundering conduct faced different bail regimes because the crimes that generated the money differed. In the special statutes whose bail conditions had previously been upheld, such as the anti-terror and organised crime laws, the stringent conditions attached to the grave offence created by that statute itself; under the PMLA they attached by reference to offences created by other statutes altogether. That mismatch, the bench held, produced anomalies that no rational nexus to the object of the Act could save.

Parliament responded through the Finance Act, 2018, which amended Section 45 with effect from 19 April 2018 and substituted the words “under this Act” for the reference to offences punishable under Part A of the Schedule. The twin conditions thereafter applied to a person accused of an offence under the PMLA itself, that is, of money laundering, aligning the trigger with the offence the statute creates. Whether that amendment cured the constitutional defect or merely relabelled it was one of the central questions in the 2022 batch, and the Court held that it cured it: the vice identified in 2017 was the classification by predicate offence, and once the conditions were tied to the laundering offence itself, the reasoning that sustained similar conditions in other special statutes applied. The parallel with the anti-terror bail regime is direct, and the architecture of the two systems, the statutory bar, the prima facie inquiry and the constitutional-court escape route, is compared in this series in the analysis of UAPA bail and judicial interpretation, which owns that comparison.

Two features of that sequence deserve to be held firmly, because public discussion routinely loses both. The first is that the 2022 Court did not overrule the 2017 analysis of the old provision; it upheld a differently worded provision and treated the defect as one of classification that the amendment had removed. The second is that the amendment which did this work was carried in a Finance Act, a Money Bill, and the validity of that route is a question the 2022 judgment expressly did not decide. The twin conditions thus stand restored on reasoning that depends on an amendment whose method of enactment remains unadjudicated, a point to which this article returns.

Why did the petitioners argue that the Act turns criminal law upside down?

Because each safeguard of the ordinary process is reversed at a different stage: the suspect must answer questions whose answers are admissible, his property is attached before trial, he is arrested on an officer’s belief, he must satisfy a court of his innocence for bail, and at trial his property is presumed tainted unless he disproves it.

That framing, pressed by the senior counsel who led the challenge, among them Kapil Sibal and Dr Abhishek Manu Singhvi, was the intellectual spine of the petitions. Its force is that it does not depend on any single provision being extreme. Compelled statements exist in customs and tax inquiries. Provisional attachment exists in several fiscal statutes. Reverse burdens exist in narcotics and corruption law. Stringent bail conditions exist in anti-terror law. What does not exist elsewhere, the petitioners argued, is the full set assembled in one statute, operating on a person who may not even know the case against him because the initiating document is withheld, and attached to an offence whose existence depends entirely on a different crime, investigated by a different agency, which may never be proved at all.

The Union’s framing was equally systematic. Money laundering, it argued, is the offence that erases the evidence of other offences; it is committed by concealment, and its victim is the financial system and, through it, the sovereignty and integrity of the state, language the statute itself uses. A legal regime for such an offence must be able to move against property at the speed at which property moves, must be able to compel the paper trail’s custodians to speak, and must place evidential burdens on the person who alone knows the money’s provenance. Proportionality, the Union submitted, is measured against that subject matter, and judged against it every provision survived.

The Vijay Madanlal Choudhary challenge: eight years to decide

The lead petition in the batch, Special Leave Petition (Criminal) No. 4634 of 2014, was filed in 2014, and petitions accumulated behind it for years as enforcement under the Act expanded. By the time the bench assembled to hear the batch, it comprised 241 petitions, writ petitions, appeals and transferred cases, brought by businessmen, companies, chartered accountants and a striking number of persons prominent in public life, including serving and former ministers and members of legislatures facing proceedings under the Act. The hearings extended over more than twenty days of argument, and the questions referred and argued covered not only the PMLA but connected provisions of other statutes whose schemes resemble it. This matters to how the judgment should be read. It is not a ruling on one arrest or one attachment; it is a systematic adjudication of a statutory scheme, delivered with the knowledge that it would govern thousands of pending investigations, which is one reason the Union pressed, and the Court accepted, that the analysis proceed provision by provision rather than by asking whether the Act felt oppressive in the aggregate.

How long was the challenge pending before the Supreme Court decided it?

The lead petition was filed in 2014, and challenges accumulated behind it for eight years before judgment on 27 July 2022. In that interval the Act was amended repeatedly, the twin conditions were struck down in 2017 and restored in amended form in 2018, and enforcement expanded across the country while the foundational questions awaited decision.

The interval also produced a doctrinal oddity. Between 2014 and 2022 the Act that the Court was examining kept changing beneath it. Amendments carried by the Finance Acts of 2015, 2016, 2018 and 2019 redefined proceeds of crime, altered the Schedule, changed the procedure for attachment and, in 2018, rewrote the bail provision the Court had struck down the previous year. A constitutional challenge to a moving statute is a challenge to several statutes in succession, and the 2022 judgment is best understood as a judgment on the Act as it stood after the 2019 amendments, with the earlier versions relevant chiefly through the Nikesh Tarachand Shah episode. Readers who treat the judgment as a verdict on the Act of 2005 misread its object.

The unsettled law the batch was assembled to settle

To see why a bench of three judges and 241 petitions were required, it helps to reconstruct how open the operative questions were in the years before the judgment. On the supply of the Enforcement Case Information Report, persons proceeded against had demanded the document in court after court, and the answers they received turned on which forum they were in and on whether the court treated the report as the equivalent of a first information report or as an internal noting; the question had no authoritative answer, and investigations proceeded under the shadow of both possibilities. On bail, the position after 2018 was genuinely confused: Nikesh Tarachand Shah had struck down the twin conditions in the form in which they applied through the Schedule, Parliament had rewritten the trigger, and accused persons arrested for laundering connected with complaints registered at different times argued, with varying success, that the struck-down provision could not be applied to them, that the amendment could not revive what the Constitution Bench reasoning had condemned, or that the amendment applied only prospectively. On statements, defence counsel across the country argued that a person summoned under Section 50 who was arrested on the strength of his own answers had in substance been compelled to incriminate himself while formally unaccused, and that the distinction between witness and accused was being manipulated by sequencing. On attachment, the reach of the power to proceed against property of equivalent value, property that was itself untainted but said to stand in place of proceeds that had been moved or spent, was contested in tribunal proceedings across the country.

The Supreme Court’s own interim interventions illustrated the unsettlement. In P. Chidambaram v Directorate of Enforcement, reported at (2019) 9 SCC 24, the Court in 2019 refused anticipatory bail to a former finance minister, taking note of the amended Section 45 and of the gravity the legislature attaches to economic offences of this class, while the foundational challenge to the amended provisions remained pending undecided. Individual cases were thus being decided, and liberty granted or refused, under a statute whose central provisions were simultaneously under an unresolved constitutional cloud. That is the context in which the 2022 judgment must be read: not as an abstract seminar on a settled statute, but as the overdue resolution of questions on which arrests, attachments and bail applications across India had been turning for the better part of a decade, and on which the courts below had been obliged to improvise.

What the Court held, provision by provision

The operative holdings can be stated shortly, and they should be stated shortly before they are analysed, because a great deal of commentary on this judgment argues with a version of it that the bench did not deliver.

On Section 3, the Court read the definition of money laundering at its full width: the offence extends to every process or activity connected with the proceeds of crime, including concealment, possession, acquisition, use, and projecting or claiming the property as untainted, and these are alternative modes of committing the one offence rather than steps in a sequence. But the Court built a limit into that width which has since done more work in trial courts than any other sentence in the judgment. The authorities cannot prosecute a person on a notional basis or on an assumption that a scheduled offence has been committed; the scheduled offence must be registered with the jurisdictional police or be pending enquiry or trial, including by way of criminal complaint before the competent forum; and if the person is finally discharged or acquitted of the scheduled offence, or the case is quashed by a competent court, there can be no offence of money laundering against him, nor against anyone claiming the property through him. The offence is standalone in the sense that a conviction for the predicate is not a precondition, but it is parasitic in the sense that a predicate case must exist and survive.

On Section 5, the Court upheld provisional attachment, emphasising the safeguards internal to the provision: the officer must have material in his possession, must record a reason to believe in writing that the property is proceeds of crime and is likely to be concealed, transferred or dealt with so as to frustrate confiscation, and must forward the order and the material to the Adjudicating Authority, which then conducts its own inquiry before confirming. Attachment is provisional, time-bound and subject to adjudication and appeal, and on that footing the Court declined to treat it as a penalty imposed without trial. On Section 8(4), the Court upheld the power to take possession of property whose attachment has been confirmed, but read a discipline of restraint into practice: possession is to be taken in exceptional circumstances, not as the automatic sequel of confirmation, a qualification that tribunals and courts have since applied to restrain the Directorate from evicting occupants of attached property while trials remain unconcluded.

On Sections 17 and 18, search and seizure and search of persons, the Court sustained the powers on the strength of their internal conditions: a recorded reason to believe, an authorisation structure, and the obligation to forward the record to the Adjudicating Authority. On Section 19, arrest, the Court upheld the power and identified its safeguards with some care: the officer must have material in his possession on the basis of which he records in writing a reason to believe that the person is guilty of an offence under the Act, the person arrested must be informed of the grounds of arrest, and must be produced before a Special Court or a magistrate within twenty-four hours, with a copy of the arrest order and the supporting material forwarded to the Adjudicating Authority in a sealed cover. The section’s conditions, the Court held, are not formalities; they are the substance of the safeguard, and non-compliance was, in later cases, to prove fatal to arrests. On Section 44, the designation of Special Courts for trial of the offence, the challenge failed without difficulty.

On Section 24, the reverse burden, the Court upheld the presumption that proceeds of crime are involved in money laundering, holding that it has a reasonable nexus with the objects of the Act and is not arbitrary, but it confined the presumption’s operation in a way its text does not spell out: the burden on the accused arises only after the prosecution establishes the foundational facts, that there was a scheduled offence, that property constituting proceeds of crime resulted from it, and that the accused is connected with that property; and what the accused must then do is rebut the presumption on a preponderance of probabilities, not prove innocence beyond reasonable doubt. On Section 45, the Court upheld the twin conditions as restored in 2018, for the classification reason already given, while recording that the provision does not oust the constitutional courts’ power to grant bail in exceptional situations, and that the bar operates at the threshold of the bail inquiry rather than as a determination of guilt.

On Section 50, the summons power, the Court’s holdings are the most consequential for daily practice. Officers of the Directorate are not police officers. A statement recorded by them under Section 50 is therefore not hit by the bar on confessions made to police, and the constitutional protection against self-incrimination, which is available to a person accused of an offence, is not attracted at the stage of a summons addressed to a person who is not then an accused. The proceedings under Section 50 are judicial proceedings for the purposes of the penal provisions on false evidence, which means the person summoned must answer truthfully on pain of prosecution for perjury, and a statement so recorded is admissible in the trial. The Court also addressed the connected complaint that the Directorate registers no first information report: the Enforcement Case Information Report, it held, is an internal document by which the agency records its decision to inquire, it is not the equivalent of a first information report under the criminal procedure code, its supply to the person under inquiry is not mandatory, and the accused’s protection lies in the material that must accompany the prosecution complaint when it is filed and in the Special Court’s power to call for the record.

What did the Court say about India’s international obligations?

The Court treated the conventions against drug trafficking and transnational organised crime and the standards of the Financial Action Task Force as the source of the statute and as relevant to its construction, accepting that a state which fails to maintain an effective anti-laundering regime bears real costs in creditworthiness, correspondent banking and trade.

That reasoning deserves one qualification the judgment itself does not emphasise. International standards require an effective offence, freezing powers and international cooperation; they do not dictate the particular domestic choices on compelled statements, bail burdens and presumptions that the petitions attacked. The step from “an effective regime is obligatory” to “this regime, in each of these particulars, is proportionate” is a step of domestic constitutional reasoning, and it is at precisely that step that the critics concentrate. The point is developed in the assessment below.

Taken as a whole, the judgment is best characterised as a validation with conditions. The conditions are real: the predicate offence must exist and survive, attachment must rest on recorded reasons and material, possession should be exceptional, arrest must comply with Section 19 in form and substance, the presumption in Section 24 waits on foundational facts, and the constitutional courts retain a residuary bail power. But the validation is also real, and broader than many expected. No provision in issue was struck down. The two holdings on which the scheme most depends in practice, the admissibility of Section 50 statements and the non-supply of the Enforcement Case Information Report, were sustained in terms. And the Court rejected the proportionality attack on the punishment under Section 4 in short order.

The trial forum: Section 44 and the shape of a laundering trial

The challenge to Section 44 failed quickly in the judgment, but the provision shapes everything the statistics record, and its design explains part of the funnel. A laundering offence is triable only by a Special Court designated under the Act for the area, and the Special Court takes cognizance of the offence on the prosecution complaint filed by the Directorate, without the committal proceedings of ordinary sessions trials. Because the offence depends on a scheduled offence tried elsewhere, the statute as amended addresses the obvious risk of two courts reaching inconsistent findings on the same underlying facts: the scheme contemplates the scheduled offence and the laundering offence being tried by the same Special Court where the predicate case is also before it, so that the foundational question, whether a crime produced proceeds, and the dependent question, whether the accused laundered them, are answered on a single record. In practice the predicate case is frequently before a different court, a magistrate, a Central Bureau of Investigation court or a state special court, and the laundering trial then waits, formally or in effect, on findings it does not control, which is the procedural mechanism behind much of the delay the Directorate’s annual report describes.

The judgment’s contribution here was negative but important: it declined to find in the forum design any constitutional infirmity, treating the Special Court mechanism as a permissible specialisation of the trial function of a kind familiar from corruption and narcotics law. The consequence is that complaints about the trial’s pace and structure must be directed where the judgment leaves them: to the administration of the Special Courts, their number, their dual charge with other statutes, and the listing practices that decide whether a laundering trial sits weekly or yearly, and to the legislature if the dependency on the predicate trial is to be re-engineered. Those are not small matters, since they determine the width of the funnel’s final gate; but after July 2022 they are matters of policy and administration rather than of constitutional law, and arguments that present them as the latter will be met with the judgment.

How the Court reasoned: the method behind the holdings

The holdings are only half the judgment; the other half is a method, and the method explains why a bench that found safeguards everywhere struck nothing down. Three features of the reasoning recur across provisions and are worth isolating, because later benches applying the judgment have worked with the method as much as with the results.

The first feature is the presumption of constitutionality operated at high strength. The Court began from the position that a statute enacted to discharge international obligations and to protect the financial system carries a heavy presumption of validity, that the legislature understands the subject it legislates on, and that a provision should be read down to save it rather than struck down where a saving construction is available. Almost every saving device in the judgment is an application of that starting point: Section 8(4) possession is saved by being confined to exceptional circumstances; Section 24 is saved by being made to wait on foundational facts; Section 19 is saved by treating its procedural conditions as mandatory substance rather than form. A reader who misses the starting presumption will misread these conditions as concessions extracted from an unwilling bench; they are better understood as the bench’s chosen instrument, the terms on which validation was granted.

The second feature is comparative validation. Again and again, a PMLA provision is sustained by placing it beside a counterpart in another special statute that the Court has previously upheld: the twin conditions beside the bail bars of the anti-terror and organised crime and narcotics statutes, whose conditions had survived challenge in decisions such as Kartar Singh v State of Punjab on the anti-terror law; the reverse burden beside presumptions in narcotics, customs and corruption law; compelled statements beside the inquiry powers of customs and income tax authorities, whose officers have never been treated as police. The comparative move does real analytical work: it converts the question “is this departure from ordinary process permissible” into the question “is this departure more extreme than departures already sustained”, and on that second question the PMLA usually wins, because its individual devices each have a pedigree somewhere in Indian statute law. The petitioners’ answer, that the combination rather than the components was the challenge, never received a comparably structured response, and the gap between the challenge as framed and the challenge as answered is the methodological heart of the later criticism.

The third feature is the treatment of the Act as a code of graded intrusions, in which the severity of each power is answered by the seniority of the officer who wields it and the judicial supervision that follows it. Attachment is answered by adjudication before the Adjudicating Authority and appeal to the Appellate Tribunal and the High Court. Arrest is answered by production before a court within twenty-four hours and by the sealed-cover transmission of the officer’s material to the Adjudicating Authority. Search is answered by recorded reasons and forwarding of records. The judgment reads the statute as a system of internal checks, and its validation is of the system so read. This is why the subsequent case law, which consists overwhelmingly of courts testing whether the internal check was actually performed in the case before them, is not a departure from the judgment but its continuation: Pankaj Bansal is Section 19’s check being enforced exactly as the 2022 reasoning described it.

The judgment in the line of special-statute doctrine

The judgment did not write on a blank page, and its most important doctrinal move becomes visible when it is placed in the line of authority on India’s special criminal statutes. For four decades the Court has sustained, with conditions, a family of departures from ordinary criminal process in statutes addressed to terrorism, organised crime, narcotics and corruption: bail bars conditioned on a prima facie view of innocence, presumptions attaching to possession or to the proceeds of specified offences, and special courts. The sustaining decisions characteristically do two things at once: they uphold the departure by reference to the gravity and distinctiveness of the subject matter, and they attach operating conditions, most often that a presumption may be drawn only after the prosecution proves the facts that trigger it, and that the accused’s rebuttal is measured on the civil standard. The locus classicus of the second move is Noor Aga v State of Punjab, decided in 2008 on the presumptions in the narcotics statute, where the Court held that the initial burden on the prosecution to establish the foundational facts is undiminished, that the presumption operates only thereafter, and that the accused discharges his burden on a preponderance of probabilities, adding that the harsher the presumption the more scrupulous the court must be in insisting on the prosecution’s first stage.

Vijay Madanlal Choudhary’s treatment of Section 24 is that tradition applied, almost term for term, to money laundering, and recognising the pedigree matters for how the holding should be argued with. When the Court held that the presumption has a reasonable nexus with the Act’s objects, it placed the provision in the sustained family; when it required foundational facts and fixed the accused’s standard at preponderance, it attached the family’s standard conditions. The review’s decision to flag the burden holding for reconsideration is therefore more radical than it may appear: it does not merely question one section of one statute, it questions, at least potentially, the terms on which the entire family of reverse-burden provisions has been sustained since Noor Aga, and a bench answering it will either reaffirm the tradition with the PMLA inside it or begin the tradition’s revision. The same is true, in the bail field, of the relationship between Section 45 as sustained and the bail bars sustained elsewhere: the 2022 judgment’s classification reasoning, that the conditions are valid because they attach to the statute’s own offence, is the reasoning of the anti-terror bail cases transposed, and the Sisodia qualification, that the bar yields to Article 21 when trial cannot conclude, has since been stated in terms the anti-terror jurisprudence will also have to absorb. The judgment is thus best understood not as an eccentricity about one financial statute but as the point at which the special-statute doctrine, assembled piecemeal over forty years, was restated as a whole for the statute that deploys more of its devices than any other, and simultaneously marked, in two particulars, as possibly due for restatement again.

The document the accused never sees

No single holding illustrates the judgment’s balance, or its cost, better than the one on the Enforcement Case Information Report, and it deserves a slower treatment than the survey above allows. In ordinary criminal process, the first information report is the founding public document of a case: it is recorded under the procedure code, a copy must be furnished to the informant and made available to the accused, and the Supreme Court has directed that reports be uploaded publicly, subject to exceptions, because the document disciplines the investigation that follows it; subsequent embellishment can be measured against the earliest version. The Enforcement Case Information Report performs the initiating function in a laundering case: it records the Directorate’s decision that material discloses a money laundering offence worth inquiring into, it is keyed to the scheduled offence on which the case depends, and everything that follows, summons, search, attachment, arrest, traces back to it. But it is created by no provision of the Act. The statute neither names it nor prescribes its form, its contents, or anyone’s right to it. It is an administrative practice, and the question before the Court was whether constitutional fair procedure fills that silence with a right of supply.

The Court held that it does not. Its reasoning proceeded in steps. The report is an internal document, the record of the agency’s own decision to inquire, and is not equivalent to a first information report, which is a statement of information about an offence given to the police and signed by the informant. The person proceeded against is not remediless: he receives the grounds of his arrest if arrested, the reasons recorded for attachment when the Adjudicating Authority issues notice in the attachment proceedings and calls on him to show cause, and the prosecution complaint with the material relied on when the case reaches the Special Court, which may itself call for the report where justice requires it. Compelled disclosure at the threshold, the Court accepted, could compromise an inquiry whose entire object is a concealed trail that disclosure itself would help to conceal further. On those steps the holding rests, and they are not frivolous steps; the staged disclosure the Court described does exist in the statute’s design.

The criticism, and it is the criticism the review notice has marked as arguable at the highest level, is that the steps do not meet at the point of maximum pressure. The stages at which disclosure does occur arrive late relative to the stages at which the accused must act. The decisive early contests under this Act are the bail application, where the accused must satisfy the court that there are reasonable grounds for believing he is not guilty, and the attachment adjudication, where he must explain his property. At the bail stage in particular, the accused is required to demonstrate the weakness of a case he has not seen: the complaint may not yet be filed, the initiating document is withheld, and the grounds of arrest state conclusions rather than the material behind them. Kapil Sibal’s argument in the review, that an accused asked to prove a negative about an undisclosed case is denied a fair opportunity to meet it, draws its force from that sequencing, and the bench that issued the review notice, which included two judges of the original bench, treated the point as one that required to be heard again. The holding stands; the reasoning that supports it has been formally reopened; and lawyers conducting bail applications under the Act frame them, in consequence, with one eye on the record as it is and one on the record as a review bench might remake it.

The offence at the centre: Section 3 and the predicate link

The Section 3 holding repays separate attention because it is the part of the judgment that trial courts apply most often and commentators discuss least. The definition of money laundering in the Act is drafted as a chain of participles: a person commits the offence who directly or indirectly attempts to indulge, knowingly assists, knowingly is a party to, or is actually involved in any process or activity connected with the proceeds of crime, including concealment, possession, acquisition or use, and projecting or claiming the property as untainted property. The petitioners attacked the definition from two directions. They argued that its breadth criminalises innocent dealing, since possession or use of property is enough, without any requirement of a laundering purpose; and they argued that the closing words, projecting or claiming as untainted, were the true core of the offence, so that the earlier participles should be read as governed by them, a reading that would have confined the offence to acts of disguise.

The Court rejected the confining reading. It treated the processes and activities named in the section as independent modes of the offence, any one of which suffices, and it read the statutory Explanation, which clarifies that the offence is a continuing one and that a person is guilty if involved in any one of the processes, as confirming that design. Possession of proceeds of crime, knowing them to be such, is therefore itself money laundering, without proof of any act of concealment or projection. Against the breadth objection, the Court located the offence’s limiting principle not in the verbs of Section 3 but in its object: everything in the section is governed by the requirement that the property be proceeds of crime, and proceeds of crime are defined by their derivation from a scheduled offence. From that anchor the Court derived the propositions that have since structured discharge litigation: the prosecution cannot proceed on an assumption or a notional basis that a scheduled offence exists; the scheduled offence must be registered or pending before the competent forum; and the laundering case cannot survive the predicate case’s termination by discharge, acquittal or quashing, because the property then ceases, in law, to be proceeds of crime at all.

The consequence is a dependency that cuts in both directions, and both directions are litigated constantly. For the accused, the collapse of the predicate is the single most powerful event in a laundering case, more valuable than any bail order, because it dissolves the offence itself rather than mitigating its process; defence strategy in laundering cases is consequently dominated by the progress of the predicate case, and accused persons routinely seek to have the laundering trial awaited, stayed or closed as the predicate moves. For the prosecution, the same dependency disciplines case selection: a laundering case built on a fragile predicate is a case built to collapse, and the Directorate’s complaints are drafted to exhibit the predicate’s registration and pendency as foundational facts, precisely as the judgment’s reading of Section 24 requires them to be proved. The 2023 refinement in Pavana Dibbur, that the person prosecuted need not himself be an accused in the scheduled offence but must be shown to have engaged in a process or activity in relation to property that is proceeds, fits into this structure without disturbing it: the link runs to the property and its criminal derivation, not necessarily to the identity of the predicate offender.

Attachment: the power that does the daily work

If Section 3 is the judgment’s most applied holding, Section 5 is its most consequential validation, because attachment is the act by which the statute touches the largest number of people, most of whom are never arrested and many of whom are never accused. The statutory scheme the Court sustained works in stages. An authorised officer who has material in his possession, on the basis of which he has reason to believe, recorded in writing, that a person possesses proceeds of crime and that the proceeds are likely to be concealed, transferred or dealt with so as to frustrate confiscation, may provisionally attach the property for a period the statute caps at one hundred and eighty days. The order and the material must be forwarded to the Adjudicating Authority, which issues notice to the person affected, hears him on the sources of his income and the character of the property, and either confirms the attachment, whereupon it continues until the trial’s conclusion, or declines to confirm it, whereupon the property is released. An appeal lies to the Appellate Tribunal and thereafter to the High Court. At the trial’s end, confirmed attachment ripens into confiscation if the Special Court finds the property involved in money laundering, and into release if it does not.

The petitioners’ attack on this scheme was that it permits the executive to inflict, on an officer’s belief, a deprivation of property indistinguishable in effect from punishment, years before any court decides whether an offence occurred, and that the confirmation proceeding before the Adjudicating Authority answers the question whether the property is proceeds of crime on material the owner may contest only after the deprivation has occurred. The Court’s answer proceeded from the scheme’s provisional and supervised character: the belief must be recorded and material-based, the deprivation is time-limited unless an independent authority confirms it after hearing the owner, the owner may appeal, and the property is not confiscated at this stage but preserved, its ownership finally determined by the trial court. To that the Court added the practical premise that runs through the whole judgment: property that is proceeds of crime, if not restrained at the stage of inquiry, will ordinarily be dissipated beyond recovery, so that a regime without provisional attachment would be a regime without a remedy against the thing the statute exists to reach.

Two qualifications within the validation have proved fertile in the tribunals. The first is the Court’s treatment of possession under Section 8(4): although the provision permits the Directorate to take possession of confirmed attached property, the Court indicated that taking possession, as opposed to restraining alienation, is justified only where exceptional circumstances are shown, and tribunals have since restrained the Directorate from taking physical possession of homes and operating premises on that footing while trials remain pending. The second is the discipline of the recorded reason: because attachment must rest on a reason to believe founded on material in possession, the attachment order and its forwarding to the Adjudicating Authority are examinable, and attachment proceedings have failed where the recorded reasons disclose no material linking the specific property to the scheduled offence, as distinct from material linking its owner to an allegation. Attachment under this Act is therefore best understood, after the judgment, as a strong power exercised through a paper discipline: it stands or falls on the contemporaneous record the officer created, and the record is where challenges to it are won and lost.

Section 50 and the compelled statement

The summons power is the point at which the statute departs most sharply from the citizen’s ordinary expectations of criminal process, and the judgment’s treatment of it repays the same slow reading. Under Section 50, an authorised officer may summon any person whose attendance he considers necessary to give evidence or produce documents, may examine that person on oath, and the proceedings are deemed to be judicial proceedings within the meaning of the provisions of the Penal Code that punish false evidence and insult to a person sitting in a judicial proceeding. The person summoned is bound to attend, bound to answer, and bound to answer truthfully; a false statement is perjury, and silence or evasion is itself punishable. The statements so recorded are then admissible in the prosecution. The petitioners’ case was that this is interrogation by another name, conducted by the investigating agency, on the subject matter of the investigation, with the answers usable to convict, and that the protection in Article 20(3) of the Constitution, that no person accused of an offence shall be compelled to be a witness against himself, is defeated if the State need only choose to summon before it arrests.

The Court’s answer turned on two connected propositions. The first is institutional: the Directorate’s officers are not police officers, a conclusion the Court supported by the accumulated authority on analogous officers, customs and excise officers having long been held not to be police officers for this purpose in the line of cases associated with Romesh Chandra Mehta v State of West Bengal and Poolpandi v Superintendent, Central Excise, with the consequence that neither the bar on confessions to police nor the procedural code governing police investigation applies to the Directorate’s inquiry as such. The second proposition is temporal: the protection against self-incrimination is, by its terms, available to a person accused of an offence, and a person summoned under Section 50 is not, at the stage of the summons, an accused in the prosecution that may later be filed; a statement recorded at that stage is therefore not the compelled testimony of an accused, and if the person is later prosecuted, the statement’s use is governed by the ordinary law of evidence, including the requirement that it have been made voluntarily in the sense the law recognises, and by the accused’s right to retract it and test it at trial.

The doctrinal criticism of this pair of propositions is the most intellectually serious criticism the judgment faces, and it should be recorded as criticism rather than adopted as conclusion. The judgment’s line between the summoned person and the accused is a formal line, drawn by the state of the file; the decisions associated with Nandini Satpathy v P. L. Dani had been understood to read the protection against self-incrimination by reference to the substance of the compulsion and the reasonable apprehension of accusation, rather than the label on the proceeding, and the criticism is that the 2022 judgment restores the label without confronting the substance: the same officer, the same file, the same questions, and arrest following within hours for answers the officer dislikes. The enforcement answer, equally serious, is administrative necessity joined to the structure of the offence: the proceeds trail exists in documents and in the knowledge of those who moved the money, the persons who can explain a layered transaction are frequently its participants, and an inquiry that cannot compel their account at the inquiry stage cannot reconstruct the trail at all, so that the alternative to compelled statements is not a gentler investigation but, in many cases, no investigation. The review notice did not flag the Section 50 holding, and it accordingly stands unreopened; but the arrest cases of 2023 and 2024, Pankaj Bansal most prominently, show the courts policing the boundary the judgment left exposed, the point at which inquiry hardens into custody, with an insistence the bare text of the 2022 judgment would not have predicted.

Bail in practice after the judgment

Because bail is where the statute touches liberty most directly, the working of Section 45 after the judgment deserves a record of its own. The twin conditions operate in sequence with the ordinary law: the prosecutor must be heard in opposition, and the court must then be satisfied of reasonable grounds for believing the accused is not guilty and will not offend on bail, a standard that directs the court to the prosecution’s own material and forbids a roving inquiry into evidence. In the years immediately after the judgment, trial courts and High Courts applied the conditions with the strictness the judgment’s validation invited, and bail in contested laundering cases was, in documented practice, difficult to obtain at the first attempt, particularly where the prosecution complaint had been filed and the court treated the complaint’s allegations as the material on which its satisfaction had to be formed.

The counterweight came, as the judgment had said it would, from the constitutional courts’ residuary power and from Article 21. The route was marked in the litigation of Manish Sisodia, a former deputy chief minister of Delhi arrested in the excise policy cases. In October 2023 the Supreme Court declined him bail on the merits of the twin conditions, while recording the prosecution’s assurance about the pace of trial and observing that if the trial did not proceed as represented, a fresh application would be open. When the trial had not materially advanced, the Court in August 2024 granted bail, holding that keeping an accused in custody for an indefinite period while a trial of great volume crawls is a violation of the right to speedy trial, that the twin conditions were not intended to convert pre-trial custody into a default, and that the principle that bail is the rule applies under this Act. In a connected line of orders through 2024, including the bail granted in the case of the Jharkhand accused Prem Prakash, the Court repeated that the maxim governing bail applies to special statutes including the PMLA, and that Section 45 must be applied case by case rather than as a formula for refusal. The position that has emerged is therefore layered: the twin conditions are valid and applicable; they are applied to the prosecution’s material at a prima facie standard; they do not require the accused to displace evidence at a trial standard; and they yield, on the authority of the Supreme Court itself, where custody becomes disproportionate to any realistic prospect of the trial concluding. A bail court that ignores any layer of that structure errs, and appellate correction of such errors has been frequent enough to constitute a doctrine of its own.

The twin conditions as a working standard

Since the bail bar decides more liberty questions than any other part of the judgment, its internal mechanics are worth one precise statement. The first condition is procedural: where the public prosecutor opposes the application, the conditions are engaged, which places on the prosecution the obligation to appear and oppose, and on the court the obligation to hear the opposition before the standard is applied. The second and third conditions are the satisfaction limbs: reasonable grounds for believing that the accused is not guilty of the offence, and that he is not likely to commit any offence while on bail. The governing gloss on the first limb comes from the cognate jurisprudence on the narcotics statute, stated in Union of India v Shiv Shanker Kesari: reasonable grounds are grounds that are substantial and probable, more than a prima facie case in the loose sense and less than proof, and the court arrives at them by a broad assessment of the material without the detailed weighing of evidence that belongs to trial, a discipline the anti-terror bail cases expressed as the bar on holding a mini-trial at the bail stage. The provision itself contains a limited proviso: a person under sixteen years of age, a woman, or a person who is sick or infirm may be released on bail if the Special Court so directs, notwithstanding the conditions, a textual acknowledgement by Parliament itself that the bar is not meant to be absolute against vulnerability.

Three features of the working standard after the 2022 judgment should be fixed, because commentary frequently garbles at least one of them. The satisfaction is formed on the prosecution’s material, the complaint and the documents relied on, together with the accused’s answer to it; the accused is not required to lead evidence or to establish innocence affirmatively. The finding is expressly provisional, recorded for the purpose of bail, and binds neither the trial nor the accused’s other remedies, which is why a refusal of bail under the conditions is not a finding of guilt and why courts granting or refusing bail repeat the formula that the observations are tentative. And the standard, though stringent, is not the end of the court’s authority: the 2022 judgment preserved the constitutional courts’ power to act in exceptional cases, and the 2024 bail decisions added the Article 21 route where custody outruns the trial’s prospects, so that the twin conditions operate as a demanding threshold within a system that retains two exits, rather than as the sealed bar they are often described as being by both the statute’s defenders, who exaggerate their necessity, and its critics, who exaggerate their absoluteness.

The criticism, at full strength

The criticism of the judgment began from inside the institution’s own recent past. Justice Rohinton Fali Nariman, who had authored Nikesh Tarachand Shah before his retirement, expressed public disappointment with the decision in a lecture delivered in March 2023, and former Chief Justice of India U. U. Lalit associated himself with the critical view later that year. When the judge who wrote the 2017 judgment and a former Chief Justice both record disagreement with a coordinate ruling, the criticism cannot be dismissed as the advocacy of disappointed litigants, and its content deserves to be stated as its authors state it.

The first strand is constitutional method. The judgment, the critics say, decides questions of liberty by deference: it moves from the seriousness of money laundering and the existence of international obligations to the validity of each challenged provision, without subjecting the individual provisions to the structured proportionality inquiry, suitability, necessity, and balancing, that the Court applies elsewhere. On this view, the holding that Section 24 bears a reasonable nexus to the Act’s objects answers an Article 14 question while the live question was under Article 21: whether a presumption of the central incriminating fact, in a statute where the accused may never have seen the initiating document, is a fair procedure at all.

The second strand concerns the Enforcement Case Information Report. The petitioners’ argument, pressed by Kapil Sibal throughout the review proceedings, is that the distinction between an internal document and a first information report answers the wrong question. Whatever the document is called, it is the record of the case the agency proposes to make; a person arrested, attached and summoned on its basis, who is denied it, defends himself against an undisclosed case, and must do so at a bail hearing at which he carries the burden of showing he is not guilty. The critics add that the judgment’s own safeguard, that the Special Court may call for the document, is a remedy the accused must seek from a court while in custody under a reversed burden.

The third strand is about Section 50. If statements compelled on oath, under threat of prosecution for falsehood, are admissible, and if the line between a person summoned and a person accused is drawn by the agency’s own choice of timing, then the protection against self-incrimination depends on the investigator’s sequencing. The criticism is sharpened by the practical reality that a summons under the Act commonly precedes arrest by hours or days: the statement is recorded while the person is formally a witness, and used after he becomes an accused. Justice Nariman’s criticism of the judgment fastened in particular on holdings of this character, which he regarded as departing from settled protections without adequate justification.

The fourth strand is empirical, and it is the strand that dominates public argument. Registrations under the Act have risen by large multiples across the past decade; the petition filed in March 2023 by fourteen political parties, led by the Indian National Congress and argued by Dr Abhishek Manu Singhvi, placed before the Court the claim that cases registered by the Directorate had risen from 209 in 2013-14 to 981 in 2020-21 and 1,180 in 2021-22, while convictions secured under the Act stood, in that account, at twenty-three in total. On the critics’ reading, a statute whose concluded trials number in the dozens after thousands of registrations, and whose attachment and custody consequences bite immediately, operates in practice as a system of sanctions without adjudication: the attachment is the punishment, the custody pending a bail application decided under a reversed burden is the punishment, and the trial, if it arrives, arrives after both have been served.

The enforcement answer, at full strength

The answer to that criticism must also be stated as its makers state it, because it is not a weak answer. The Directorate’s first point is structural: a money laundering prosecution cannot be faster than the predicate case it depends on. The Directorate does not investigate the scheduled offences; police forces, the Central Bureau of Investigation, state anti-corruption bureaus and other agencies do, and the laundering case must wait on theirs for the foundational fact that a crime produced proceeds. Delay in the predicate is not delay of the Directorate’s making, and the judgment’s own holding, that acquittal or discharge in the predicate ends the laundering case, builds that dependency into the law. A conviction count for the PMLA is therefore partly a report on the speed of every other investigative agency and of the ordinary trial courts.

The second point concerns the nature of the harm. Proceeds of crime that are not frozen during investigation are, in the ordinary case, not frozen at all: they are layered through accounts and jurisdictions within days. Provisional attachment, on this view, is not a shortcut to punishment but the only form the remedy can take, and it is hedged by adjudication before an authority, appeal to a tribunal and the supervisory jurisdiction of the High Courts, with arrest, remand, bail and trial each supervised by courts at successive stages. The Directorate’s actions are, as the Union submitted, judicially reviewable at multiple points, and are in fact reviewed constantly; bail orders, attachment releases and quashed arrests fill the reports.

The third point is the political composition of the caseload, and on this the Directorate has put a figure in the public record. In data published by the agency for the period ending 31 January 2023, of 5,906 Enforcement Case Information Reports recorded since the Act commenced, 176 concerned serving or former members of Parliament, members of state legislatures and members of legislative councils: 2.98 per cent of the total. The agency’s case, stated by its officers and by the Solicitor General in court, is that a caseload more than ninety-seven per cent of which consists of bank frauds, corporate defaults, narcotics proceeds, cybercrime and corruption cases unconnected with party politics cannot be characterised by the small fraction that involves political figures, however prominently those cases are reported.

The counter-figure, and it is a genuinely contested figure rather than a refuted one, was put to the Supreme Court by Dr Singhvi in the fourteen-party petition: that of 121 political leaders investigated by the Directorate, ninety-five per cent belonged to parties in opposition. The two figures are not measuring the same thing, and the difference between them is instructive. The agency’s 2.98 per cent divides political cases by all cases; the petitioners’ ninety-five per cent divides opposition politicians by political cases only. Both can be simultaneously accurate. What the data establishes is that political cases are a small share of the caseload and that, within that share, opposition figures predominate overwhelmingly in the period the petition examined. What it does not establish, on either side, is motive: predominance is consistent with targeting, and equally consistent with the fact that the predicate corruption cases of a decade are brought disproportionately against those who held office in it. When the fourteen-party petition asked the Court to convert the pattern into prospective guidelines on arrest and remand, a bench of Chief Justice D. Y. Chandrachud and Justice J. B. Pardiwala declined on 5 April 2023, observing that political leaders stand on the same footing as other citizens and enjoy no higher immunity, that general guidelines laid down without the facts of a case would be dangerous, and that a person aggrieved in an individual case has the courts open to him, as the petition was withdrawn. The political question was returned, expressly, to the political arena.

This article follows that allocation. It does not adjudicate the allegation of political targeting, because the material that would decide it, the internal file notings of individual cases, is not in the public record, and because both the pattern and its innocent explanation are genuinely available on the published figures. What can be adjudicated, and is, is the arithmetic of the conviction debate, because there the record is complete enough to settle the question that is actually being argued about.

The denominator table: both conviction rates are true

No statistic in Indian legal discourse is quoted more confidently and understood less than the conviction rate under this Act. The Directorate and the government quote a figure above ninety per cent. Critics quote a figure that rounds to nothing. Each side treats the other’s number as a fabrication. Neither is a fabrication; they are quotients with different denominators, and once the denominators are laid side by side the argument dissolves into a more precise and more troubling finding than either side states. This is the third time this series has had to perform the same operation: the identical confusion structures the debate about anti-terror prosecutions, resolved by separating registrations from concluded trials in the anti-terror law enforcement record, and the debate about insolvency recoveries, resolved by separating admitted claims from admitted assets in the IBC implementation record. The rule those articles established applies here without modification: an enforcement statistic is meaningless until its denominator is named, and the interesting question is always what happened to the cases the denominator excludes.

The table is built on the Directorate’s own published data for the position at 31 January 2023, the release from which both famous percentages derive, with the one stage that release does not report, the framing of charges, taken from a Finance Ministry answer in the Rajya Sabha of 29 July 2025, whose different period is stated in its row and must not be merged with the others.

Denominator in public use	Figure and period	Convictions measured against it	The resulting rate	Who quotes it	What it does and does not measure
Cases registered (Enforcement Case Information Reports recorded since the Act commenced on 1 July 2005)	5,906 cases to 31 January 2023, Directorate of Enforcement published data	24 cases ending in conviction	About 0.4 per cent	Opposition parties and critics of the agency	Measures the throughput of the whole system across seventeen and a half years. It does not measure the merits of any case, and it counts as failures thousands of cases that have not reached a verdict either way
Prosecution complaints filed before the Special Courts	1,142 complaints to 31 January 2023, Directorate of Enforcement published data	24 cases ending in conviction	About 2.1 per cent	Critics, and answers given in Parliament	Measures how many filed prosecutions have yet produced a conviction. Since most complaints are still awaiting trial, the figure mostly measures delay rather than failure
Cases reaching the trial stage (charges framed)	Charges framed in 300 of 5,892 cases taken up between 1 January 2015 and 30 June 2025, Finance Ministry answer in the Rajya Sabha, 29 July 2025	8 conviction orders in that period, covering 15 persons	Charges framed in about 5 per cent of cases taken up	Parliamentary debate on the Act’s working	Measures the width of the gate between complaint and trial. It belongs to a different period from the other rows and is shown to identify the stage at which cases stall, not to be averaged with them
Trials concluded on merits	25 trials completed to 31 January 2023, Directorate of Enforcement published data	24 convictions and 1 acquittal	96 per cent	The Directorate and the government	Measures the quality of the small set of cases that survived to judgment, in which 45 accused persons were convicted. It excludes every pending case, which is to say nearly the entire system

Read down the table and the controversy resolves itself. The government’s figure is the last row read upward: of the trials that reached a verdict, twenty-four of twenty-five ended in conviction, and on updated figures released with the Directorate’s annual report for 2024-25, of 47 cases decided since the Act commenced, 44 ended in conviction and three in acquittal, a rate of 93.6 per cent which the Director of Enforcement, Rahul Navin, defended publicly in May 2025 while conceding that several investigations had been pending for a very long time. The critics’ figure is the first row read downward: of 5,906 cases registered over the Act’s life to January 2023, twenty-four had produced a conviction, which is 0.4 per cent; stated against the persons arrested in the same data, 513, or against the complaints filed, 1,142, the picture is of the same order. Both computations are arithmetically correct. Neither is responsive to the question the public debate is actually about.

The question the debate is actually about is not the ratio in either direction. It is the shape of the funnel between the first row and the last. In over two decades of operation, counting the Act’s life from its commencement in 2005, the statute has produced a few dozen concluded trials against several thousand registrations: 7,771 cases registered and 1,739 prosecution complaints filed by the time of the 2024-25 annual report, against 47 decided cases. Everything that the statute does to people, attachment of homes and accounts, arrest, custody measured in months while bail is contested under a reversed burden, the professional and political consequences of being named in a prosecution complaint, happens in the part of the funnel where verdicts have not arrived and, on the demonstrated pace of the Special Courts, mostly will not arrive for years. The honest finding is therefore the one neither side leads with: conviction is not what this statute principally does. Process is what it does, and the judgment of July 2022 is a judgment about the legality of that process, which is why its terms matter far more than the conviction arithmetic that dominates discussion of it.

Two disciplines follow for any reader citing these numbers, and they are the same disciplines this series applied to the anti-terror and insolvency records. First, never quote a percentage from this table without its row: a conviction rate stated without a denominator is not a statistic but a slogan. Second, never merge periods: the figures move every year, the agency’s releases and the ministry’s parliamentary answers are snapshots of a running system, and a table that averages snapshots taken in different years manufactures a precision the record does not contain. The figures given here are tied to their sources and dates precisely so that a reader can replace them, row by row, as later releases appear, without disturbing the structure, which will survive every update: a wide mouth, a narrow gate at the framing of charges, a very small number of verdicts, and a very high conviction share among those few.

Who is actually inside the funnel

The public image of the Act’s caseload is formed by its most visible cases, and the denominator table corrects that image in one further respect worth stating. The Directorate’s published data for the period to 31 January 2023 records 513 arrests across the entire life of the Act to that date, against 5,906 cases registered: fewer than one case in eleven involved an arrest at all. The overwhelming majority of the caseload consists of financial investigations arising from bank and corporate fraud, corruption cases transmitted from other agencies, narcotics and contraband proceeds, and cyber-enabled fraud, in which the operative events are the recording of the case, the summonses, the attachment of property and the eventual complaint, not custody. The statute’s most famous defendants, ministers, legislators, senior officials and prominent businessmen, are unrepresentative of its ordinary subjects, who are directors of failed companies, bank officers, intermediaries, accountants and the holders of properties through which money is said to have passed, many of them proceeded against in the attachment track without ever being arrested or, in some instances, without being accused of the offence at all, since attachment runs against property and may reach persons who hold it for others.

That distribution matters to the evaluation of the judgment in two ways. First, it qualifies the custodial critique: the bail jurisprudence that dominates commentary concerns a minority of the caseload, while the majority experience of the Act is proprietary, the freezing of assets and accounts for years on end, which is why the Section 8(4) possession qualification and the paper discipline of attachment, rather than the twin conditions, are the parts of the judgment with the widest practical reach. Second, it explains the judgment’s structure: a bench asked to rule on a statute whose daily operation is administrative and proprietary, but whose constitutional challenge was framed in the language of personal liberty by petitioners many of whom faced custody, produced a judgment that validates the proprietary machinery in the language of administrative safeguards and reserves its most anxious passages for the points where property process converts into personal coercion: arrest, statement and bail. Reading the judgment with the caseload’s true distribution in mind makes its emphases intelligible, and makes the review’s selection of its two issues, the secret initiating document and the reversed burden, the two devices that govern the proprietary majority as much as the custodial minority, appear well chosen.

A note on the figures used in this article

Because this article’s central exhibit is a table of contested numbers, its method of assembly should be on the record. Every figure in the table and the surrounding analysis is drawn from one of three named sources: the data published by the Directorate of Enforcement for the position at 31 January 2023, which is the release from which the agency’s own 2.98 per cent figure for cases against legislators and its ninety-six per cent conviction figure both come; the Finance Ministry’s written answer in the Rajya Sabha of 29 July 2025, given by the Minister of State for Finance, Pankaj Chaudhary, covering cases taken up between 1 January 2015 and 30 June 2025; and the Directorate’s annual report for 2024-25 together with the Director’s public presentation of it in May 2025. No figure has been carried from one period into another, no two sources have been averaged, and where two sources give different totals for overlapping ground, the difference is a difference of period or of counting unit, cases, complaints, persons, orders, and is identified as such in the text. Figures quoted from litigation, the fourteen political parties’ claim of twenty-three convictions and of ninety-five per cent opposition share among political leaders investigated, are attributed to the petitions that advanced them and are used only as claims in the debate, not as findings of this article. A reader who wishes to test any number in the table should be able to find it in the named source for the named period; that, and not the direction of the conclusion, is what distinguishes a denominator from a slogan.

Why trials do not conclude: the Directorate’s own account

Before treating the funnel’s shape as an accusation, a reader should have the explanation offered by the institution that operates it, because the Directorate has published one. In its annual report for 2024-25, the first the agency has issued in that form, the Directorate recorded one hundred Special Courts designated under the Act across the country and attributed the slow completion of trials to a combination of causes it itemised: the structural dependence of the laundering trial on the predicate trial, so that delay in the predicate case is transmitted into the laundering case; the complexity of the financial material, involving layered structures, large volumes of records and frequently cross-border transactions requiring forensic examination; the burden on Special Courts that are also charged with cases under other statutes; and the interruption of trials by interlocutory applications, writ petitions and bail proceedings, some of which travel to the High Courts and the Supreme Court and suspend the trial while they are decided. The Director of Enforcement, speaking to the report in May 2025, acknowledged candidly that several investigations had been pending for a very long time and that completing them and filing final prosecution complaints would be a focus of the agency’s work.

That account is largely consistent with the observable record, and an honest assessment should say so: predicate dependency is a design feature this article has already traced to the statute’s structure and to the 2022 judgment’s own reading of Section 3, and no one familiar with Indian trial courts will doubt the interlocutory burden. But the account also has a boundary, and the boundary matters. The causes the Directorate lists explain why trials, once begun, are slow; they do not fully explain the width of the gate before trial, the stage at which, on the Finance Ministry’s figures for 2015 to 2025, charges had been framed in only about five per cent of cases taken up, since the framing of charges depends principally on the filing and readiness of the prosecution complaint itself. Nor do they touch the question the critics press hardest, which is not why the machine runs slowly but why it is set running so much more often than it is allowed to finish: registrations in the thousands are a choice of enforcement policy, not a product of judicial delay. Both halves of that analysis can be true together, and the denominator table is the instrument that keeps them separate.

The judgment its own court flagged within a month

Eight review petitions were filed against the judgment, the lead one by Karti P. Chidambaram, a member of Parliament who is himself an accused under the Act and whose petition is Review Petition (Criminal) No. 219 of 2022. On 25 August 2022, a bench presided over by Chief Justice N. V. Ramana, sitting with Justices Dinesh Maheshwari and C. T. Ravikumar, two of the three judges who had delivered the judgment under challenge, issued notice on the reviews. The bench recorded that prima facie at least two conclusions of the judgment required reconsideration: the conclusion that a copy of the Enforcement Case Information Report need not be supplied to the accused, and the conclusion upholding the reversal of the burden of proof and the presumption it creates against the accused. The bench allowed the unusual course of an open court hearing, review petitions ordinarily being decided in chambers, and made clear that the notice was confined to those aspects and was not a general reopening of the judgment.

The significance of that order is easy to understate and should not be. A notice on review is not a reversal, and the two flagged conclusions remain the law unless and until a bench holds otherwise. But a three-judge bench that included two authors of the judgment looked at its own holdings within four weeks and identified the two that carried the heaviest constitutional weight, the secrecy of the initiating document and the reversal of the presumption of innocence, as the two that might not stand. The judgment therefore occupies an unusual position in the reports: it is the governing authority on every question it decided, it is cited daily by trial courts and by the Directorate, and its two most consequential holdings are under a subsisting notice of reconsideration issued by the Court itself.

What has happened since is, in its own way, as informative as the notice. The reviews were not listed for substantive hearing for nearly a year after notice. They were then listed repeatedly before a bench of Justices Surya Kant, C. T. Ravikumar and Ujjal Bhuyan and adjourned repeatedly, at the request of the parties and on one occasion at the request of the Solicitor General, who sought time to prepare. Justice Ravikumar’s retirement required the bench to be reconstituted, with Justice N. Kotiswar Singh joining it, and the matter continued to be listed and adjourned without a concluded hearing. In separate proceedings, other benches declined to convert the pendency of the review into a wider unsettlement: the Court refused requests to refer the judgment to a larger bench in collateral challenges and observed that accused persons in other cases cannot take a piggyback ride on the review petitions, whose scope the notice had confined to two aspects. The position, stated precisely, is that the review is admitted on two issues, has been allowed an open court hearing, has passed through successively reconstituted benches, and remains pending and undecided. Any account that describes the judgment of July 2022 as the final word on the Act is, on the Court’s own record, wrong; any account that describes it as overruled or suspended is equally wrong. It is law under a stated reservation, and the reservation is the Court’s.

For the practitioner, the consequence of that position is concrete. The two flagged holdings may be argued against in a way that no other part of the judgment can be: counsel appearing in a bail application or an attachment appeal can submit, accurately, that the supply of the Enforcement Case Information Report and the operation of Section 24 are questions the Supreme Court has itself marked as open, and can invite the court seised of the matter to take that into account in the exercise of its own discretion. The remaining holdings, on attachment, search, arrest, statements and bail, are not open in that sense, and arguments against them must be built as distinctions or as invitations to apply the judgment’s internal conditions strictly, which is in practice how they have been built since 2022. Researchers and reporters assembling a record of a case in this area have to preserve that distinction document by document, which is why those who cover enforcement litigation keep their case files, correspondence and source material in structured private workspaces; the working methods described in how journalists use VaultBook to protect sources are built for exactly that discipline of keeping a record that can be produced, in order, years later.

The international frame, and what it can and cannot carry

Because the judgment leans on India’s international obligations at several points, the frame deserves to be set out accurately rather than invoked. The Prevention of Money Laundering Act was Parliament’s response to a sequence of commitments: the United Nations Convention Against Illicit Traffic in Narcotic Drugs and Psychotropic Substances of 1988, which first obliged states to criminalise the laundering of drug proceeds; the United Nations Convention Against Transnational Organized Crime of 2000, concluded at Palermo, which generalised the obligation to the proceeds of serious crime; and the standards of the Financial Action Task Force, the intergovernmental body whose recommendations, though formally soft law, are enforced in practice by mutual evaluation and by the consequences for a state’s financial standing of a poor assessment. India became a full member of the Financial Action Task Force in 2010. The statute’s unusual features, its standalone offence built on a predicate, its provisional attachment, its reverse burden, were defended before the Court as the domestic machinery of that international framework, and the Court accepted the framework as a legitimate and weighty interpretive context, noting the costs a state bears, in creditworthiness, correspondent banking relationships and the ease of trade, when its anti-laundering regime is assessed as ineffective.

Two observations discipline the use of that frame, one available to each side. For the Union, the frame’s authority is not merely rhetorical: the Financial Action Task Force’s mutual evaluation of India, adopted in 2024, placed India in regular follow-up, the strongest of its assessment categories, a result only a small number of states achieve, and that standing is a national asset in the financial system which a judgment dismantling the enforcement machinery would have put at risk. For the critics, the frame proves less than the judgment asks of it, for two reasons. The international instruments require an offence, freezing and confiscation powers and cooperation; they do not prescribe compelled admissible statements, presumptions against the accused, or bail conditions, which are choices of domestic law to be justified by domestic constitutional standards. And the Financial Action Task Force’s own methodology measures an anti-laundering regime by effectiveness, prominently including whether investigations lead to prosecutions and convictions in numbers proportionate to the offending; an evaluation framework that asks that question is an uncomfortable authority for a system whose concluded trials number in the dozens, and the 2024 evaluation itself recorded the need for India to improve the completion of money laundering prosecutions relative to the scale of investigation. The frame therefore cuts in both directions, and the judgment’s use of it is best read as establishing the legitimacy of the statute’s object rather than as discharging the proportionality inquiry for each of its means, which is where the critics, including the review petitioners, continue to press.

The road not taken: what a strike-down would have meant

A judgment is illuminated by the consequences of the decision it declined to give, and in this case those consequences were concrete enough to have formed part of the atmosphere of the hearing. Had the bench struck down the twin conditions again, or the attachment power, or the admissibility of Section 50 statements, the ruling would have operated across thousands of pending investigations and prosecutions at every stage: attachments supporting pending complaints would have been reopened, custody granted or refused under a condemned provision would have required rehearing, statements forming the spine of filed complaints would have been vulnerable, and the Directorate would have faced the reconstruction of its standard practice under the constraint, stated by a later bench in refusing a reference to a larger bench, that a wholesale unsettlement would bring pending investigations to a halt. The interregnum after Nikesh Tarachand Shah supplied a small-scale preview: the 2017 judgment had directed that bail matters decided under the struck-down provision be reheard, and courts across the country spent months re-deciding custody under a provision that Parliament then restored in amended form within five months, leaving a cohort of cases whose bail history ran through three regimes in a year.

None of that is an argument that unconstitutional provisions should be sustained for convenience, and the Court did not put it in those terms. It is, however, the context in which the bench’s chosen technique, validation by condition, should be understood. Faced with a decade-old, fully operational enforcement system and a challenge to its every component, the Court did what constitutional courts frequently do with such systems: it sustained the architecture and constitutionalised its operation, converting the safeguards the Union had described as features of the scheme into conditions of its validity. The price of that technique is the one the review notice identifies: conditions attached in reasoning are only as strong as the willingness of later courts to enforce them, and the two conditions the Court attached to the system’s fairness at its most sensitive points, the accused’s knowledge of the case and the prosecution’s burden of proof, are the two its own notice has marked as not finally settled. The road not taken and the road taken thus converge on the same unfinished questions, which is perhaps the most accurate single statement that can be made about this judgment’s place in the law.

The question the judgment refused to answer: the Money Bill

A second reservation in the judgment is less known than the review and may prove larger. Many of the amendments that reshaped the Act between 2015 and 2019, including the 2018 amendment that restored the twin conditions after Nikesh Tarachand Shah, were enacted not as ordinary legislation but as parts of Finance Acts certified as Money Bills, a route under Article 110 of the Constitution that excludes the Rajya Sabha from effective participation. The petitioners argued that amendments to a criminal statute creating offences, powers of arrest and presumptions are not provisions on taxation or expenditure within Article 110, and that their passage as Money Bills was unconstitutional; and they pointed out, as Dr Singhvi emphasised in argument, that the consequence of success would not be marginal: the judgment itself recorded that if the Money Bill challenge were correct, the amendments carried by that route would fall.

The bench declined to decide the question. In Rojer Mathew v South Indian Bank, decided in November 2019, a five-judge bench had doubted the reasoning by which the Aadhaar Act had been sustained as a Money Bill and had referred the interpretation of Article 110 to a bench of seven judges. The Vijay Madanlal bench held that the Money Bill challenge to the PMLA amendments would abide by that reference, and decided the case on the text as enacted. The seven-judge bench was constituted for procedural directions in October 2023, when Chief Justice D. Y. Chandrachud indicated that a call on hearing the reference would be taken; Solicitor General Tushar Mehta opposed prioritising it. No substantive decision on the reference has been delivered, and the question accordingly remains where the 2022 judgment left it: undecided, by design, pending before a larger bench. The constitutional mechanics of the Money Bill route, and what would follow from each possible answer, are owned by the article on the money bill and finance bill distinction in this series.

Sit with the resulting position for a moment, because it is genuinely strange and entirely accurate. A statute was upheld in 2022 against comprehensive challenge. Two of its holdings were flagged for reconsideration within a month. And the amendments on which its most litigated provision, the twin conditions, depend for their present form were enacted by a method whose validity the upholding judgment expressly left to another bench that has not yet answered. None of that makes the judgment any less binding. It makes the description of the judgment as a settlement of the law unsustainable, and it means that the most important PMLA decisions of the coming years may be a review about two holdings and a reference about a procedure, neither of which questions whether money laundering should be an offence at all.

The sequel: the safeguards line of 2023 and 2024

The judgment’s internal conditions, which could have remained decorative, did not. In the two years after July 2022 a line of decisions by other benches enforced the safeguards the judgment had identified, and in one respect added to them, with the combined effect that the working law of arrest and bail under the Act is measurably more protective than the bare text of the 2022 holdings suggests. This section states that line in outline; the detail of individual prosecutions belongs elsewhere.

The leading case is Pankaj Bansal v Union of India, decided on 3 October 2023 by a bench of Justices A. S. Bopanna and P. V. Sanjay Kumar. Section 19 obliges the arresting officer to inform the arrested person of the grounds of arrest, and the practice had grown of communicating them orally or by showing them without furnishing a copy. The Court held that to give true meaning to the constitutional and statutory mandate, a copy of the written grounds of arrest must be furnished to the arrested person as a matter of course and without exception; the arrests before it, of two directors of a real estate group, were held unsustainable, and the remand orders founded on them fell with them. The same judgment held that mere non-cooperation with a summons under Section 50 does not, without more, furnish a ground for arrest under Section 19, since an investigating officer cannot proceed on the footing that the person summoned is bound to confess. Within weeks, a bench of Justices Bela M. Trivedi and Satish Chandra Sharma, in Ram Kishor Arora v Directorate of Enforcement in December 2023, qualified the reach of Pankaj Bansal: it was held not to operate retrospectively, and oral communication of the grounds at the time of arrest, followed by furnishing the written grounds within twenty-four hours, was treated as satisfying Section 19 and Article 22(1) of the Constitution. The coexistence of the two rulings has left the precise timing rule contested between High Courts, and the safest statement of the position is that written grounds must reach the arrested person promptly and that arrests have repeatedly been undone for failures in this requirement.

On the property side, the corresponding discipline came in Pavana Dibbur v Directorate of Enforcement, decided on 29 November 2023, which addressed who may be prosecuted and in respect of what property. The Court held that a person who is not an accused in the scheduled offence may nonetheless be prosecuted under the Act if he knowingly assists in or is involved in a process or activity connected with the proceeds of crime, but that the prosecution must still anchor itself in property that is in fact proceeds of a scheduled offence, and that mere possession of property acquired before the criminal activity, without involvement in the laundering process, does not constitute the offence. Stated shortly, the offence requires both criminal property and a laundering act in relation to it; neither alone suffices. That reading, consistent with the predicate-dependency the 2022 judgment had built into Section 3, has become the standard framing in discharge applications.

On bail, the movement ran the other way from what a reader of Section 45 alone would predict. In Arvind Kejriwal v Directorate of Enforcement, decided on 12 July 2024, a bench of Justices Sanjiv Khanna and Dipankar Datta upheld the arrest before it as satisfying Section 19 on its facts, holding that the officer’s reasons to believe must rest on all relevant material, including material that absolves the person, and may not be a selective compilation; but the bench granted interim bail and referred to a larger bench the question whether a need or necessity to arrest must be read into Section 19 as a condition of a valid arrest, a question it expressly connected with the doctrine of proportionality, and which, like the review and the Money Bill reference, remains undecided. In Manish Sisodia v Directorate of Enforcement, decided on 9 August 2024, a bench of Justices B. R. Gavai and K. V. Viswanathan granted bail to an accused who had spent about seventeen months in custody across the connected cases, holding that the twin conditions do not displace the right under Article 21 to a speedy trial, and that where the trial cannot conclude within a reasonable time and the accused has already undergone a substantial period of custody, bail follows notwithstanding the statutory bar, a conclusion the Court grounded in the principle that bail is the rule and jail the exception, which it held applies to prosecutions under this Act as it does elsewhere.

A further decision of the same period restructured the relationship between arrest and complaint. In Tarsem Lal v Directorate of Enforcement, decided on 16 May 2024 by a bench of Justices Abhay S. Oka and Ujjal Bhuyan, the Court held that the power of arrest under Section 19 is a power exercisable during investigation, and that once a prosecution complaint has been filed and the Special Court has taken cognizance of it, the Directorate cannot thereafter arrest the accused in that case; and that an accused who was not arrested during the investigation, and who appears before the Special Court on summons, does not have to run the gauntlet of the twin conditions merely to remain at liberty, the Special Court being empowered instead to take a bond for appearance under the procedure code. The ruling matters beyond its facts because it removed a widespread practice under which arrest after complaint, followed by a bail application decided under Section 45, had become a routine stage of laundering prosecutions, and it did so by reading the 2022 judgment’s own premise, that arrest is an investigative act hedged by investigative safeguards, to its conclusion.

The general position described in this article is legal information and analysis; it is not advice on any individual case, and a person facing proceedings under the Act needs advice on his own facts, because the limitation periods, remand rules and filing deadlines in this statute are unforgiving of delay. What the sequel establishes at the level of doctrine can nevertheless be stated plainly. Vijay Madanlal Choudhary fixed the validity of the scheme. The decisions that followed it have been occupied almost entirely with the conditions the judgment itself attached to that validity: recorded reasons, written grounds, complete material, exceptional possession, foundational facts, and a bail bar that yields when detention outlasts any prospect of trial. Whether that pattern is a stable equilibrium or the slow reconstruction of the statute by safeguards is the live interpretive question, and the answer will be given, if it is given, by the pending review and the pending references rather than by fresh challenges to provisions the Court has already upheld.

What remains arguable: the practitioner’s map

For the advocate, the judgment’s practical content is a map of which arguments are closed, which are open, and which run through the judgment’s own conditions. The closed arguments should be identified first, because courts treat attempts to reopen them as concluded by authority: the constitutional validity of Sections 3, 5, 17, 18, 19, 24, 44, 45 and 50 as schemes; the proposition that Directorate officers are not police officers; the admissibility in principle of Section 50 statements; and the validity of the twin conditions as restored in 2018. A submission that any of these is unconstitutional, addressed to a trial court or a High Court, will fail, and competent advocacy does not spend its credit there.

The open arguments are the ones this article has tracked. Two holdings stand under the Supreme Court’s own notice of reconsideration: the non-supply of the Enforcement Case Information Report and the presumption under Section 24. One question stands referred to a larger bench and undecided: whether the Finance Act amendments required the ordinary legislative route rather than the Money Bill route, with the consequence, recorded in the judgment itself, that the amendments would fall if the challenge succeeded. A second question stands referred and undecided: whether necessity to arrest is a condition of a valid arrest under Section 19, as the Kejriwal bench framed it in July 2024. In each of these four matters, an advocate may accurately submit that the point is open at the highest level, and may invite the court before him to exercise whatever discretion it has, on bail, on the pace of trial, on the restraint of possession, in a manner that does not foreclose the accused’s position if the larger questions are answered in his favour.

Between the closed and the open lies the third and busiest category: the arguments that run through the judgment’s internal conditions. Was the reason to believe recorded in writing, on material in possession, before the attachment or arrest it justifies? Was the material complete, including what absolves, as the Kejriwal judgment requires? Were written grounds of arrest furnished, and when? Has the prosecution established the foundational facts that alone trigger the presumption? Is the scheduled offence registered and surviving, and is the specific property linked to it, or only the accused’s reputation? Is possession of attached property being sought without the exceptional circumstances the judgment requires? Is custody continuing past any realistic prospect of trial, so that Article 21 overrides the bail bar on the Sisodia principle? These are the questions on which laundering cases are in fact won and lost after 2022, and they share a common character: none of them asks a court to depart from Vijay Madanlal Choudhary, and all of them ask a court to enforce it.

The general rule for reading enforcement statistics

The series thesis thread can be stated in its general form, because this is the third occasion on which it has been needed. The anti-terror enforcement record showed a conviction rate near ninety per cent for the specialised agency that counts concluded trials and a rate in low single figures for the system that counts registrations. The insolvency recovery record showed a recovery percentage that changed by hundreds of points depending on whether the denominator was claims admitted, assets realised, or the value creditors themselves assigned at admission. This article’s table shows the same phenomenon in its purest form, a rate of ninety-six per cent and a rate of under one per cent computed from the same agency’s same release. The rule that the three records jointly establish is that an Indian enforcement statistic is a statement about a chosen stage of a funnel, and that the political work is always done by the choice of stage: governments quote the stage at which the state wins, oppositions quote the stage at which the citizen is caught, and the stage neither quotes, the width of the gate between complaint and verdict, is where the system’s real character is recorded. A reader equipped with that rule does not need this table again; he needs only to ask, of any figure offered to him, which row it is.

What the record settles, and what it does not

A closing verdict can be stated with the discipline the subject requires. The judgment of 27 July 2022 settled the constitutional validity of the Enforcement Directorate’s core powers under the Prevention of Money Laundering Act, 2002: attachment, search, arrest, compelled statements, the reverse burden and the twin conditions for bail are all lawful parts of the statute, and that settlement is not in suspense. The judgment did not settle, because it was not asked to settle in a form it could answer, whether the statute is well administered; and the enforcement statistics, read with their denominators restored, settle a narrower and harder proposition in place of the slogans traded about them. In the life of the Act, verdicts are rare, convictions among verdicts are the norm, and the consequences that matter to the people inside the system are imposed by process long before any verdict. A reader who understands those three sentences understands the PMLA debate better than most of its participants.

What remains open is a specific and enumerable list, and it is the measure of this judgment’s unusual afterlife that the list was written partly by the Court itself: whether the Enforcement Case Information Report must be supplied to the accused; whether the presumption in Section 24 can stand in its upheld form; whether amendments to a criminal statute may be carried by Money Bill; and whether an arrest under Section 19 must be necessary as well as grounded. Until benches answer those questions, Vijay Madanlal Choudhary is best cited as what it is: a comprehensive validation of a statute, delivered with conditions, two of whose holdings its own court marked for a second look within a month of delivering them.

Frequently Asked Questions

Q: What did Vijay Madanlal Choudhary uphold?

The judgment of 27 July 2022 upheld the constitutional validity of the Prevention of Money Laundering Act’s operative machinery: the wide definition of the offence in Section 3, provisional attachment under Section 5, taking possession under Section 8(4), search and seizure under Sections 17 and 18, arrest under Section 19, the reverse burden under Section 24, trial by Special Court under Section 44, the twin conditions for bail under Section 45, and the summons and statement power under Section 50. It also held that the Enforcement Case Information Report is an internal document whose supply to the accused is not mandatory. No provision in issue was struck down, though the Court attached conditions to several holdings, including that possession of attached property should be exceptional and that the presumption in Section 24 operates only after the prosecution proves the foundational facts.

Q: Why are statements to Enforcement Directorate officers admissible?

Because the Supreme Court held that officers of the Directorate are not police officers, so the rules that exclude confessions made to police and restrict the use of statements recorded by police do not apply to them. A statement recorded under Section 50 of the Act is taken in proceedings deemed to be judicial proceedings, the person summoned is bound to answer truthfully, and the constitutional protection against self-incrimination was held not to be attracted at the summons stage because the person summoned is not at that point an accused in a prosecution. The statement can therefore be used in evidence at the trial, subject to the ordinary law of evidence and to the accused’s right to retract it and challenge it. The holding follows the older line of authority on customs and excise officers, who have likewise been held not to be police officers for this purpose.

Q: What are the twin conditions for bail under the PMLA?

The twin conditions are the two additional requirements in Section 45 of the Prevention of Money Laundering Act that a court must satisfy before releasing a person accused under the Act on bail. First, the public prosecutor must be given an opportunity to oppose the application. Second, where the application is opposed, the court must be satisfied that there are reasonable grounds for believing that the accused is not guilty of the offence and that he is not likely to commit any offence while on bail. These conditions operate in addition to the ordinary considerations under the criminal procedure code, and they invert the usual bail inquiry, because the court must record a preliminary satisfaction about innocence rather than asking only about flight risk, tampering and the prima facie case. A limited proviso permits a Special Court to release a person under sixteen, a woman, or a person who is sick or infirm notwithstanding the conditions.

Q: Why were the twin conditions struck down and then revived?

In Nikesh Tarachand Shah v Union of India, decided on 23 November 2017, the Supreme Court struck the conditions down as manifestly arbitrary and violative of Articles 14 and 21 of the Constitution, because they applied by reference to the punishment attached to the predicate offence in Part A of the Schedule, so that identical laundering conduct attracted different bail regimes depending on which crime had generated the money, a classification with no rational connection to the laundering itself. Parliament then amended Section 45 through the Finance Act, 2018, with effect from 19 April 2018, substituting the words “under this Act” so that the conditions attach to the offence of money laundering itself. In Vijay Madanlal Choudhary in 2022 the Court upheld the restored conditions, holding that the amendment had cured the classification defect identified in 2017 and that the conditions were comparable to bail bars sustained in other special statutes.

Q: Is the Vijay Madanlal judgment under review?

Yes, in a limited and precisely defined sense. Eight review petitions were filed, the lead one by Karti P. Chidambaram, and on 25 August 2022 a bench presided over by Chief Justice N. V. Ramana issued notice, recording that prima facie at least two conclusions required reconsideration: the holding that a copy of the Enforcement Case Information Report need not be supplied to the accused, and the holding upholding the reversal of the burden of proof under Section 24. The bench allowed an open court hearing, which is unusual for review petitions. The petitions were thereafter listed repeatedly before successively reconstituted benches and adjourned, and they remain pending and undecided. The notice is not a suspension or a reversal: the judgment remains governing law on every point, including the two flagged holdings, unless and until a bench holds otherwise.

Q: What is the money bill question in the PMLA challenge?

Many amendments that reshaped the Prevention of Money Laundering Act between 2015 and 2019, including the 2018 amendment that restored the twin bail conditions after they were struck down, were enacted as parts of Finance Acts certified as Money Bills under Article 110 of the Constitution, a procedure that excludes the Rajya Sabha from effective participation. The petitioners argued that amendments to a criminal statute dealing with arrest, attachment and presumptions are not taxation or expenditure provisions within Article 110 and could not validly travel by that route. The bench in Vijay Madanlal Choudhary expressly declined to decide the question, because a five-judge bench in Rojer Mathew v South Indian Bank had already referred the interpretation of Article 110 to a seven-judge bench. That reference remains undecided, so the validity of the Money Bill route for these amendments is still an open question, and the 2022 judgment itself recorded that if the challenge succeeded the amendments carried by that route would fall.

Q: How many PMLA trials have actually been completed?

Very few, and the exact figure depends on the source and the date, which must always be stated with it. On the Directorate of Enforcement’s published data for the position at 31 January 2023, trials had been completed in 25 cases since the Act commenced in 2005, producing convictions in 24 and one acquittal, with 45 accused persons convicted. On the updated figures released with the Directorate’s annual report for 2024-25, 47 cases had been decided in the Act’s history, 44 ending in conviction and three in acquittal. Against those completed trials stand 5,906 cases registered to January 2023 and 7,771 registered by the time of the 2024-25 report. The small number of concluded trials, rather than the outcome of those trials, is the central enforcement fact about the Act, because it means almost every consequence the statute imposes is imposed before any verdict.

Q: Why do both sides quote different PMLA conviction rates?

Because each side divides the same convictions by a different denominator, and both divisions are arithmetically correct. The Directorate and the government divide convictions by trials concluded on merits: on the agency’s data to 31 January 2023, 24 convictions from 25 completed trials, a rate of ninety-six per cent, updated to 93.6 per cent on the 2024-25 figures of 44 convictions from 47 decided cases. Critics divide the same convictions by cases registered: 24 convictions against 5,906 registrations is about 0.4 per cent. The first figure measures the quality of the few cases that reached a verdict and ignores the thousands still pending; the second measures the throughput of the whole system and treats every pending case as a failure. Neither figure is fabricated. The informative number is the one neither side leads with: the small count of concluded trials against several thousand registrations, which shows that the process, not the verdict, is the statute’s main product.

Q: Did the Supreme Court require the ED to give the accused a copy of the ECIR?

No. The Court held that the Enforcement Case Information Report is an internal document by which the Directorate records its decision to open an inquiry, that it is not the equivalent of a first information report under the criminal procedure code, and that supplying it to the person under inquiry is not mandatory. The Court reasoned that the accused is not left without knowledge of the case: he receives the grounds of arrest if arrested, the recorded reasons when the Adjudicating Authority issues notice in attachment proceedings, and the prosecution complaint with the material relied on when the case reaches the Special Court, which may itself call for the report. This holding is one of the two that the Supreme Court flagged for reconsideration when it issued notice on the review petitions on 25 August 2022, so it remains the law but under a subsisting notice that it may be revisited.

Q: What did the Supreme Court hold about the definition of money laundering in Section 3?

The Court read the definition at its full width and located its limit in the property rather than in the acts. Section 3 covers every process or activity connected with the proceeds of crime, including concealment, possession, acquisition, use and projecting or claiming the property as untainted, and the Court held that these are alternative modes of committing the one offence, so that knowing possession of proceeds of crime is itself money laundering without any separate act of disguise. The limiting principle is that the property must actually be proceeds of crime derived from a scheduled offence: the prosecution cannot proceed on a notional assumption that a scheduled offence exists, the scheduled offence must be registered or pending before the competent forum, and if the accused is finally discharged or acquitted in the scheduled offence, or that case is quashed, the money laundering case cannot survive against him or against anyone claiming the property through him.

Q: What safeguards did the Supreme Court find in the power of arrest under Section 19?

The Court upheld the power of arrest but treated its statutory conditions as substantive safeguards rather than formalities. The arresting officer must have material in his possession, must record in writing a reason to believe, based on that material, that the person is guilty of an offence under the Act, must inform the arrested person of the grounds of arrest, and must produce him before a Special Court or a magistrate within twenty-four hours, forwarding a copy of the arrest order and the material to the Adjudicating Authority in a sealed cover. Because the validation of the section rested on those conditions, later courts have enforced them strictly: in Pankaj Bansal in October 2023 the Supreme Court held that a copy of the written grounds of arrest must be furnished to the arrested person as a matter of course, and arrests have repeatedly been held unsustainable where the Section 19 conditions were not complied with.

Q: Can the ED take possession of attached property before the trial is over?

The statute permits it in a limited form, and the Supreme Court narrowed the practice. Under Section 8(4), once the Adjudicating Authority confirms a provisional attachment, the Directorate may take possession of the property, which can therefore happen years before the trial concludes. Upholding the provision in Vijay Madanlal Choudhary, the Court indicated that taking possession, as distinct from restraining the property’s transfer, should occur only in exceptional circumstances and not as the automatic consequence of confirmation. Appellate tribunals have since applied that qualification to restrain the Directorate from taking physical possession of homes and business premises while trials remain pending, directing that the position on the ground be preserved instead. The distinction that matters in practice is between attachment, which freezes dealings in the property and is routine, and possession, which displaces the occupant and which the judgment confines to exceptional cases.

Q: Why did the Supreme Court uphold the reverse burden in Section 24?

Section 24 requires a court to presume that proceeds of crime are involved in money laundering unless the contrary is proved. The Court upheld the presumption on the ground that it bears a reasonable nexus with the objects of the Act, reasoning that the provenance of concealed property lies peculiarly within the knowledge of the person who holds it, so that requiring him to explain it, once the prosecution has done its part, is a proportionate design. The qualification is essential to the holding: the presumption does not operate at large. The prosecution must first establish the foundational facts, that a scheduled offence occurred, that property constituting its proceeds resulted, and that the accused is connected with that property; only then does the burden shift, and the accused discharges it on a preponderance of probabilities rather than by proving innocence beyond reasonable doubt. This holding is the second of the two flagged for reconsideration in the review notice of 25 August 2022.

Q: Which judges decided Vijay Madanlal Choudhary and how many petitions were before them?

The judgment was delivered on 27 July 2022 by a three-judge bench of the Supreme Court comprising Justice A. M. Khanwilkar, Justice Dinesh Maheshwari and Justice C. T. Ravikumar, with the judgment authored by Justice Khanwilkar, who retired shortly afterwards. The bench disposed of a batch of 241 petitions, writ petitions, appeals and transferred cases challenging the Prevention of Money Laundering Act, the lead petition having been filed in 2014 and challenges having accumulated behind it for eight years. The petitioners included businessmen, companies and a number of persons prominent in public life facing proceedings under the Act, and the hearings extended over more than twenty days of argument. The judgment is reported at (2023) 12 SCC 1. Its scale explains both its authority and the significance of the review notice issued against it within a month, since a ruling of that breadth governs thousands of pending investigations across the country.

Q: What did Justice Nariman, who struck down the twin conditions, say about the 2022 judgment?

Justice Rohinton Fali Nariman, who authored the 2017 judgment in Nikesh Tarachand Shah striking down the twin conditions before his retirement from the Supreme Court, expressed public disappointment with the Vijay Madanlal Choudhary judgment in a lecture delivered in March 2023, criticising its treatment of the protections the earlier case law had maintained, including the admissibility of compelled statements and the approach to the reversal of the burden of proof. Former Chief Justice of India U. U. Lalit associated himself with the critical view later in 2023. Their interventions matter because criticism of a judgment by the judge who wrote the decision it displaced, and by a former head of the institution, carries a weight that ordinary commentary does not, and the substance of their objections substantially overlaps the two issues the Court itself flagged for reconsideration when it issued notice on the review petitions: the secrecy of the Enforcement Case Information Report and the presumption under Section 24.

Q: What did the Supreme Court decide in Pankaj Bansal about the grounds of arrest?

In Pankaj Bansal v Union of India, decided on 3 October 2023 by a bench of Justices A. S. Bopanna and P. V. Sanjay Kumar, the Supreme Court held that the requirement in Section 19 of the Prevention of Money Laundering Act to inform an arrested person of the grounds of arrest means that a copy of the written grounds must be furnished to him as a matter of course and without exception; oral communication or merely showing the grounds is not enough. The arrests before the Court were held unsustainable for non-compliance, and the remand orders founded on them fell. The same judgment held that mere non-cooperation with a summons under Section 50 is not by itself a ground for arrest under Section 19. In December 2023 another bench, in Ram Kishor Arora, held that Pankaj Bansal does not operate retrospectively and that furnishing written grounds within twenty-four hours of arrest satisfies the requirement, leaving the precise timing rule debated between High Courts.

Q: Can an accused get bail under the PMLA when the trial is delayed for years?

Yes, on the authority of the Supreme Court itself, notwithstanding the twin conditions. The conditions in Section 45 are valid and applicable, but they do not displace the right to a speedy trial under Article 21 of the Constitution. In Manish Sisodia v Directorate of Enforcement, decided on 9 August 2024, the Court granted bail to an accused who had spent about seventeen months in custody, holding that where a trial of great volume shows no prospect of concluding within a reasonable time and the accused has already undergone substantial custody, bail follows, and that the principle that bail is the rule and jail the exception applies to prosecutions under this Act. The Court had earlier declined the same accused bail in October 2023 while recording the prosecution’s position on the pace of trial, and granted it when the trial had not advanced as represented. The twin conditions therefore govern the threshold of a bail application, but indefinite pre-trial detention is not a lawful outcome under them.

Q: What is the question about the necessity of arrest that a larger bench must still decide?

In Arvind Kejriwal v Directorate of Enforcement, decided on 12 July 2024, a bench of Justices Sanjiv Khanna and Dipankar Datta upheld the arrest before it as satisfying Section 19 of the Prevention of Money Laundering Act, holding that the officer’s reasons to believe must rest on all relevant material, including material that absolves the person, and may not be a selective compilation. But the bench granted interim bail and referred to a larger bench the further question whether a need or necessity to arrest must be read into Section 19 as a condition of a valid arrest, so that an arrest could be invalid not only because the belief behind it was defective but because arrest itself was unnecessary on the facts, for instance where the accused was cooperating and the investigation could proceed without custody. The bench expressly connected the question with the doctrine of proportionality. The reference remains undecided, so the present position is that Section 19 requires recorded reasons to believe based on complete material, and whether it also requires demonstrated necessity is an open question at the highest level.

Q: What happens to the PMLA amendments if the larger bench rejects the Money Bill route?

The consequences would be large, and the Vijay Madanlal bench itself said so. The amendments carried through Finance Acts between 2015 and 2019 include the 2018 amendment that restored the twin bail conditions after Nikesh Tarachand Shah, changes to the definition of proceeds of crime and to the attachment procedure, and expansions of the Schedule. If the seven-judge bench in the Rojer Mathew reference were to hold that such amendments could not validly be enacted as Money Bills under Article 110, the amendments carried by that route would, on the reasoning the 2022 judgment recorded, be liable to be declared ineffective and unconstitutional, and the Act would fall back to its earlier text in those respects, including a bail provision in the form struck down in 2017. Parliament could respond by re-enacting the amendments through the ordinary procedure with the participation of both Houses. Until the reference is decided, the amendments stand and operate, and every court in the country applies them.

Q: What share of the ED’s cases involve politicians, and why is the figure disputed?

The dispute is the denominator problem in political form. In data published by the Directorate of Enforcement for the period ending 31 January 2023, 176 of the 5,906 Enforcement Case Information Reports recorded since the Act commenced concerned serving or former members of Parliament and state legislatures: 2.98 per cent of the caseload. In the petition filed in the Supreme Court in March 2023 by fourteen political parties, senior counsel Dr Abhishek Manu Singhvi submitted that of 121 political leaders investigated by the Directorate, ninety-five per cent belonged to parties in opposition. The figures are not contradictory, because they divide by different totals: the agency divides political cases by all cases, the petitioners divided opposition politicians by political cases alone. Both can be accurate at once. What the data does not settle is motive, since a predominance of opposition figures among political cases is consistent both with selective targeting and with the pattern of predicate corruption investigations, which are brought disproportionately against those who previously held office.