For fourteen years a foreign power killed people inside Pakistan almost every week, and Pakistan’s government publicly objected to every single death while privately clearing many of them. The United States flew remotely piloted aircraft over the mountains of the northwest, watched suspected militants through cameras controlled from a desert base in Nevada, and fired missiles that landed in compounds, on vehicles, at funerals, and occasionally on the wrong house. By the time the last Hellfire missile struck the tribal belt, the Central Intelligence Agency had conducted more than four hundred strikes and ended somewhere between two and a half thousand and four thousand lives. The aerial campaign over the Federally Administered Tribal Areas became the most extensive program of cross-border killing any state had run in the modern era, and it unfolded on the soil of a nominal American ally.

That campaign matters far beyond its own grim arithmetic. It established, in plain view of the world, that the territory of a nuclear-armed state could be struck repeatedly by an outside power without the relationship collapsing, without a war, and without consequences severe enough to make the striking power stop. Washington proved that Pakistan was penetrable, that Pakistan’s protests could be absorbed, and that the political cost of killing terrorists on Pakistani ground was survivable. A decade later, when unknown gunmen began eliminating India’s most wanted on the same soil, the precedent was already set. New Delhi did not copy the American method. It studied what the American method revealed and chose a quieter instrument. To understand why India reaches for motorcycles and pistols rather than missiles and aircraft, it helps first to understand the program that came before, what it accomplished, what it destroyed, and why it eventually ground to a halt.

Background and Triggers

The American program in Pakistan was born from a problem of geography and a problem of will. After the attacks of September 11, 2001, the United States invaded Afghanistan and drove al-Qaeda’s leadership out of its training camps and into flight. Many of the surviving fighters, including senior planners, did not flee far. They crossed the porous and mountainous frontier into the Pakistani tribal districts, where the writ of Islamabad had always been thin and where the Pashtun code of hospitality offered shelter to armed guests. Within months of the fall of Kabul, Washington faced a frustrating reality. The men it most wanted to kill or capture had relocated to the territory of a country that was officially cooperating with the war on terror, yet whose military and intelligence service would not, or could not, root them out.

Pakistan in this period occupied a contradictory position. President Pervez Musharraf had pledged cooperation after a blunt set of demands from the Bush administration, and his government allowed American overflights, shared intelligence on certain targets, and arrested a number of al-Qaeda figures in its cities. At the same time, the Pakistani security establishment retained deep ties to militant networks it regarded as strategic assets, and the army was reluctant to send large formations into the tribal belt to fight a counterinsurgency that would be bloody, unpopular, and politically toxic. The result was a vacuum. The fighters were there, the Pakistani state would not clear them out, and a ground operation by American forces would have meant invading an ally. Into that vacuum stepped a piece of technology that had been maturing for a decade.

It helps to understand why the tribal belt was so hospitable to fugitives in the first place. The seven agencies of the Federally Administered Tribal Areas were governed under a colonial-era legal code, the Frontier Crimes Regulations, that the British had imposed and that independent Pakistan had never replaced. Under that code the agencies had no ordinary courts, no provincial assembly representation of the normal kind, and a system of collective tribal responsibility administered by a political agent rather than by the institutions that governed the rest of the country. Pakistani law, in the fullest sense, did not run there. The army rarely entered. The result was a strip of territory the size of a small country where the state was a rumor and where an armed man with the protection of a host clan could live for years undisturbed. Foreign fighters who arrived after 2001 did not have to build sanctuary from nothing. They moved into a sanctuary that the structure of the Pakistani state had already created and maintained for generations.

Washington examined its options in those first years after 2001 and found every conventional path effectively blocked. A full-scale American ground operation in the tribal belt would have meant a second war, this one waged on the territory of a declared ally, carrying all the diplomatic rupture and military risk that such an act implied. Pressuring Islamabad to send its own army into the agencies produced, for years, only half-measures, negotiated truces that collapsed within months, and clearing operations that emptied a valley only to watch the militants filter back once the soldiers withdrew. Diplomacy by itself could not reach men who answered to no government and held no territory a treaty could bind. Special operations raids by American forces carried the permanent danger of a captured or killed serviceman deep inside Pakistan, a political catastrophe that no administration wanted to risk more than once. Economic pressure on Islamabad was constrained by the fact that the United States needed Pakistani cooperation on supply routes, on intelligence sharing, and on the larger Afghan war. Set against that field of unattractive choices, the armed aircraft looked less like one option among many and more like the only instrument that promised measurable results without an unbearable political price. That sense of having no acceptable alternative is essential to understanding why a tool that began as a narrow expedient grew, across a decade, into the centerpiece of American counterterrorism. It was not adopted because anyone judged it clearly right. It was adopted because every other path had been weighed and found worse, and because the technology happened to mature at the precise moment the strategic problem became acute. A program born of that logic was always going to be difficult to halt, because halting it would have required finding the good alternative that had never existed in the first place.

The armed Predator had a long gestation. The aircraft began life in the 1990s as a surveillance platform, an unmanned machine that could loiter over a battlefield for many hours and beam back video. Engineers at General Atomics and within the Air Force had experimented through the late 1990s with arming the airframe, fitting it with the laser-guided Hellfire missile originally designed to be fired from helicopters against tanks. By the time of the September attacks, the marriage of a long-endurance surveillance aircraft and a precision missile had been tested but barely used. Afghanistan became its proving ground. The first known lethal use of an armed Predator came in the opening weeks of the Afghan war, and over the following two years the Agency and the military refined the tactics, the intelligence pipelines, and the legal authorities that would govern killing by remote control.

Several features of the technology made it uniquely suited to the problem Washington faced. An armed Predator could stay airborne for the better part of a day, which meant it could watch a compound, follow a vehicle, and wait for the moment when a target was isolated. It carried a camera good enough to distinguish individuals and a missile small enough to strike a single building rather than level a block. It risked no American pilot, which removed the political nightmare of a captured or killed serviceman. And because it was operated by an intelligence agency under cover of classification, its use could be neither confirmed nor denied, which suited a campaign that both governments needed to keep officially invisible. No other instrument in the American arsenal combined persistence, precision, deniability, and zero risk to American personnel. The aircraft did not merely make the campaign possible. It made the campaign tempting in a way that a manned bomber or a commando raid never could.

What turned an Afghan tool into a Pakistani program was a single target and a single negotiation. By 2004 a tribal militant named Nek Muhammad Wazir had become a serious problem for the Pakistani army in South Waziristan. He had fought the military to a humiliating standstill, signed a peace deal that embarrassed the generals, and then resumed his insurgency. Pakistan wanted him dead but did not want to be seen killing him, and it did not have the means to strike him cleanly. The United States wanted access to the tribal areas for its own hunt. The two needs met in a quiet arrangement. On June 18, 2004, an American missile killed Nek Muhammad and several others near Wana. Pakistan’s army publicly claimed the strike as its own work, a fiction that fooled few but established the template for everything that followed. The foreign power would do the killing, the host government would absorb or deflect the attribution, and the program would live in a permanent fog of official denial.

That first strike set in motion a campaign that would run, with varying intensity, until 2018. It is worth noting at the outset that the program was never formally declared, never debated in advance by the United States Congress, and never acknowledged by Pakistan as something it had agreed to. The campaign existed in the space between what governments do and what governments admit. That ambiguity was not an accident or an oversight. It was the central design feature, and it shaped every dimension of how the strikes were conducted, counted, justified, and eventually wound down. The global legal architecture around killing on foreign soil had not caught up to the technology, and the United States exploited that gap for more than a decade.

The First Strike and the Bush-Era Program

The campaign that began with Nek Muhammad moved cautiously for its first four years. Under President George W. Bush the United States conducted a relatively small number of strikes in Pakistan, somewhere in the range of four to five dozen in total across roughly four and a half years. The pace was deliberate. The Agency was still building the intelligence networks that made target identification possible, the political relationship with Islamabad was delicate, and the technology itself was being scaled up. Each strike in this early period tended to be aimed at a named, senior figure rather than at a pattern of suspicious activity, and the targets were heavily concentrated in North and South Waziristan, the two tribal agencies that had become the dense core of the militant presence.

Building the intelligence picture took years and a particular kind of human infrastructure. A missile is only as good as the information that aims it, and in a territory where Americans could not walk the streets, that information had to come from somewhere else. It came from a combination of sources, signals intelligence intercepted from phones and radios, imagery from the loitering aircraft themselves, material shared by Pakistani intelligence, and, crucially, a network of paid local informants recruited to identify houses, mark vehicles, and confirm the presence of a target. That last category was the most fragile and the most dangerous, both for the informants, who were hunted and killed by militants when exposed, and for the integrity of the program, because an informant with a private grudge could in principle point an American missile at a personal enemy. The early Bush years were spent assembling this apparatus, and the modest strike count of the period reflects how slowly a reliable targeting pipeline could be built.

Several strikes from the Bush years became notorious and revealed, early, the moral hazards that would dog the program for its entire life. In May 2005 a missile near the Afghan border killed Haitham al-Yemeni, a bomb-maker the Agency had been hunting. Later that same year a strike at Miranshah killed Abu Hamza Rabia, described in reporting at the time as al-Qaeda’s third-ranking figure, along with several others. The Rabia strike was held up by supporters of the program as proof of concept. A senior planner had been removed without an American soldier setting foot on the ground and without a public confrontation with Pakistan. Critics noted something else about that same operation, that earlier attempts on Rabia had killed members of his family, including children, and that the precision the program advertised was real only some of the time.

Two strikes in Bajaur agency in 2006 made the cost of imprecision impossible to ignore. In January a missile attack on the village of Damadola was aimed at Ayman al-Zawahiri, the Egyptian physician who served as al-Qaeda’s deputy and chief ideologue. Zawahiri was not there. Eighteen people died, among them women and children, and the strike produced furious protests across Pakistan and a diplomatic headache for both governments. In October a second Bajaur operation destroyed a religious seminary at Chenagai, and the death toll in that single strike has been estimated at roughly eighty. Pakistani authorities described the dead as militants in training; local accounts and human rights researchers disputed that, arguing that the seminary held boys and young men who were not combatants. The truth in such cases was almost always contested, almost always unverifiable from the outside, and almost always buried under competing claims. That pattern of irresolvable dispute over who exactly had died would become one of the defining features of the entire fourteen-year record.

A strike’s aftermath, in the Bush years, also revealed how the campaign would handle its own mistakes, which was to say it would not handle them at all in any public sense. There was no acknowledgment, no investigation released to the public, no compensation process visible to the outside world. A missile landed, a building collapsed, a number of people died, accounts conflicted, and within days the news cycle moved on while the Agency neither confirmed the strike nor offered an account of it. The pattern of silence served the campaign’s deniability, but it also meant that every error compounded into grievance with nothing to discharge it. A family that lost a member to a strike had no forum, no admission, no apology, and no recourse. The accumulated weight of that unaddressed grievance would become, over the following decade, one of the strongest arguments the program’s critics could make.

A feature of the program worth dwelling on, because it shaped everything downstream, is how little formal political authorization underpinned it. The United States Congress never held a dedicated debate on whether the country should run a sustained killing campaign inside Pakistan, never voted the program up or down, and for years was briefed on it only in the restricted settings reserved for covert action. The legal foundation rested largely on the broad authorization for the use of military force that Congress had passed in the days after the September 2001 attacks, a resolution aimed at the perpetrators of those attacks and the forces that had harbored them. Lawyers in successive administrations stretched that authorization to cover an expanding list of groups and a widening geography, including organizations and territory that had no connection to the original 2001 plot. Whether the resolution could legitimately bear that weight was contested by legal scholars throughout, but the contest remained largely academic, because the program was classified, the key documents were secret, and no court was positioned to rule. The result was a lethal effort of extraordinary scope operating on a thin and elastic legal base, sustained less by clear democratic mandate than by executive interpretation, bureaucratic momentum, and the absence of any institution willing or able to force a reckoning. This matters for the historical assessment because it means the program was never, in any meaningful sense, chosen by the American public or its legislature. It was chosen by a small circle of officials, defended in classified settings, and revealed to the public only in fragments, through journalism and leaks and the occasional speech, long after the practice was entrenched. An effort built that way is difficult to stop, because there is no single vote that created it and therefore no single vote that can end it. It fades, when it fades, for practical reasons rather than democratic ones, and that is, in the end, close to how the Pakistan program came to a close.

By the time Bush left office in January 2009, the program had a settled shape but a modest scale. It killed a handful of senior figures, it generated periodic outrage, and it operated under the polite fiction that Pakistan was not a party to it. Roughly fifty strikes had been carried out. The program was, in the language of bureaucracy, a tool in regular use but not yet a campaign at industrial tempo. That was about to change, and the change would come from a president who had campaigned as an opponent of the previous administration’s wars.

The Obama Escalation

Barack Obama entered office having promised to wind down the war in Iraq and to refocus American power on the fight against al-Qaeda. He did not regard the drone campaign as one of his predecessor’s excesses. He regarded it as the centerpiece of a smarter, lighter, less costly way of waging that fight, an alternative to the large occupying armies that had bled the United States in Iraq and Afghanistan. Within days of his inauguration he authorized strikes in Pakistan, and over the next several years he expanded the program dramatically. Under Obama the campaign moved from periodic operation to sustained pressure. The year 2010 saw the peak, with well over a hundred strikes recorded in a single twelve-month period, more than double the entire output of the Bush years compressed into one year of the new administration.

What drove the surge was a combination of strategic conviction and institutional confidence. The new administration believed that al-Qaeda’s core in the tribal belt represented an active and continuing danger to the American homeland, and it concluded that relentless pressure, rather than periodic strikes, was the way to dismantle it. The intelligence apparatus assembled over the Bush years had also matured. There were now more aircraft, more analysts, more informants, and a faster pipeline running from the identification of a target to the firing of a missile. A campaign that had been constrained by the slow construction of its own machinery was suddenly able to operate that machinery at full speed. The escalation, in other words, was not only a policy choice. It was also the moment when a system that had been under construction for five years finally came fully online.

Escalation was not only a matter of volume. It involved a change in the kind of target the Agency was willing to hit. In the Bush era most operations had been what analysts called personality strikes, attacks on a specific named individual whose identity had been established through intelligence. Under Obama the Agency expanded its use of what came to be called signature strikes. A signature strike did not require the operators to know the name of the person they were about to kill. It required only that the people on the ground exhibited a pattern of behavior, a signature, that intelligence analysts associated with militancy. Military-age men moving in a particular way, gathering at a particular kind of compound, handling what looked like weapons, traveling a particular route. The expansion of signature strikes is the single most important reason the program’s tempo rose so steeply, and it is also the single most important reason the civilian death toll became so contested. When you kill people whose names you do not know, you cannot be certain afterward exactly who they were.

A signature strike also rested on a counting practice that made the program’s casualty figures look better than they were. Reporting indicated that the United States, in its internal accounting, treated all military-age males killed in a strike zone as presumptive combatants unless posthumous intelligence proved them innocent. That presumption guaranteed, almost as a matter of arithmetic, that the campaign’s official civilian toll would stay low, because the burden of proof ran the wrong way. A young man killed in a strike was counted as a militant by default, and the dead cannot supply evidence of their own innocence. Independent trackers and human rights researchers rejected this counting method, and the gap between the program’s internal figures and the figures produced by outside organizations is, in large part, the gap between two definitions of who counts as a civilian.

Behind the rising strike count sat a bureaucratic apparatus that the Obama administration built to manage the killing, and the existence of that apparatus tells its own story about how the program changed. Reporting described an internal process, sometimes referred to as a disposition matrix, through which names were nominated, reviewed, and approved for lethal action, and through which the president himself, in certain categories of case, signed off on who would be struck. Regular meetings inside the national security bureaucracy reviewed the roster of targets. Government lawyers vetted the legal theory for each category of strike. An organization had been constructed to make the act of killing routine, repeatable, and defensible, and the very smoothness of that machinery troubled some of the officials who operated it. A process designed to impose discipline on lethal decisions also made those decisions easier to take, because a name that has passed through a formal review feels less like a moral choice and more like the output of a system. International criticism mounted as the machinery became visible. United Nations human rights officials questioned the legality of the strikes and the secrecy surrounding the casualty figures. European allies grew uneasy about intelligence cooperation that might implicate them in killings they could not legally have carried out themselves. Human rights organizations documented individual strikes, named individual civilian dead, and pressed for an accounting that never fully arrived. The administration answered with the argument that the program was lawful, precise, and a humane alternative to broader war, and it was that argument, more than any change of practice, that the 2013 National Defense University speech was crafted to deliver. The apparatus, in short, had two faces. One face was the disciplined review process that the administration presented as evidence of restraint. The other face was the same process seen from outside, a quiet, classified system for the regular production of death, accountable to almost no one beyond the officials who ran it.

Geography of the campaign remained remarkably stable through the escalation. The overwhelming majority of strikes landed in North Waziristan, with a smaller but significant share in South Waziristan and scattered operations in other tribal agencies such as Kurram and Orakzai. North Waziristan in particular became the program’s center of gravity because it had become the center of gravity for the militants themselves, a sanctuary that sheltered the Haqqani network, foreign al-Qaeda fighters, and factions of the Pakistani Taliban. The town of Miranshah and the surrounding Datta Khel area appear again and again in the strike record. Anyone who wishes to understand the human texture of the campaign should study the tribal terrain of North and South Waziristan, because the program cannot be separated from the specific place it was waged, a landscape of stone compounds, narrow valleys, and a population that had no vote in any of the decisions made about it.

A strike in March 2011 at Datta Khel showed how badly the signature method could fail. A missile attack on a gathering in the Datta Khel area killed a large number of people, and local accounts and subsequent reporting indicated that the gathering was a jirga, a tribal assembly convened to settle a dispute over a chromite mine. Elders, not fighters, made up much of the assembly. The strike became one of the most cited cases in the argument that the campaign’s pattern-of-behavior targeting could not reliably tell a war council from a peace council, because both look, from twenty thousand feet, like a group of armed men sitting together. The Datta Khel jirga strike did more to harden Pakistani opinion against the program than almost any other single attack, and it occurred in the same year that the broader relationship between the two governments would collapse.

Escalation also produced the program’s most strategically consequential kill. In August 2009 a missile struck a compound in South Waziristan and killed Baitullah Mehsud, the founder and leader of the Tehrik-i-Taliban Pakistan, the umbrella movement waging war against the Pakistani state itself. Baitullah had been responsible, by most accounts, for a wave of suicide bombings inside Pakistan and was widely blamed for orchestrating the assassination of former prime minister Benazir Bhutto. His death was one of the rare moments when American and Pakistani interests aligned with perfect clarity, because the man the Agency killed was an enemy of Islamabad as much as an enemy of Washington. The Baitullah strike points to a complication that ran through the whole program. The campaign targeted at least three different categories of militant, the al-Qaeda figures who threatened the American homeland, the Afghan-focused fighters who threatened American troops across the border, and the Pakistani Taliban who threatened the Pakistani state, and Pakistan’s attitude toward the strikes shifted depending on which category was being hit.

The Numbers: Strikes, Casualties, and the Geography of Death

Any honest account of the program must reckon with the fact that no one knows precisely how many strikes occurred or precisely how many people died, and that the absence of precise knowledge was itself a product of how the campaign was designed. The Agency did not publish a strike log. Pakistan did not publish a casualty count for its own tribal districts. Journalists were largely barred from the war zone, and the few who reached it did so at considerable risk. What exists instead is a set of estimates compiled by independent organizations from media reports, official statements, leaked documents, and field research, and those estimates diverge.

The most widely cited tally comes from the Bureau of Investigative Journalism, a London-based organization that maintained a detailed dataset of the campaign. Its figures, covering 2004 through 2018, record at least four hundred and thirty confirmed strikes inside Pakistan. The Bureau estimated total deaths in a range of roughly two thousand five hundred to four thousand. Within that total it estimated civilian deaths in a range of several hundred to nearly a thousand, and it estimated that somewhere between one hundred seventy and two hundred of the dead were children. Other trackers produced different numbers. The Long War Journal, which used a narrower definition of who counted as a civilian, consistently recorded lower noncombatant tolls and argued that the great majority of the dead were fighters. The New America Foundation produced figures that generally fell between the two. The honest position is that the range itself is the finding. The campaign killed thousands, a meaningful fraction of them were not combatants, and the exact accounting will never be settled because the system was never built to permit a settlement.

It is worth pausing on why three serious organizations, looking at the same war, produced three different sets of numbers. Part of the divergence was definitional. The trackers disagreed about who counted as a civilian, with some treating any unidentified military-age male as a probable militant and others treating an unidentified person as a civilian until proven otherwise. Part of it was sourcing. The trackers drew on Pakistani newspapers, international wire services, official statements, and field research, and those sources frequently contradicted one another, with a militant group, a Pakistani official, and a local resident describing the same strike in incompatible terms. Part of it was the simple fog of a closed conflict, where a strike in a remote valley might be reported days later by a single stringer working from secondhand accounts. The lesson is not that one tracker was honest and the others were not. The lesson is that a campaign conducted in deliberate secrecy makes precise accounting structurally impossible, and that anyone who cites a single confident figure for the program’s death toll is either guessing or selling.

Pakistan was the largest but not the only theater where this kind of warfare unfolded, and placing it beside the others sharpens what was distinctive about it. Over the same broad period the United States conducted parallel campaigns of airstrikes and covert action in Yemen and in Somalia, and it continued lethal operations in Afghanistan itself, where the presence of American ground troops gave those strikes a different legal and political character. Among all the places where the United States killed by remote control outside a conventional declared battlefield, Pakistan absorbed the heaviest concentration of strikes and one of the heaviest tolls. What set the Pakistan theater apart was not only its scale but its relationship to the host state. Yemen’s government, for stretches of the period, openly cooperated and at times took public credit for strikes American aircraft had carried out. Somalia had, for much of the period, no functioning central government in a position to consent or to object. Pakistan was different. It was a substantial, nuclear-armed state with a powerful army, a functioning if fragile government, and a population that watched the strikes with fury, and it maintained, for fourteen years, the fiction that it had not agreed to any of it. The Pakistan campaign therefore became the hardest case, the test of whether a major sovereign state could be struck repeatedly while both governments preserved a shared public lie. That the arrangement held for as long as it did, and that it ultimately frayed rather than exploded, is one of the most studied features of the whole era. Analysts of cross-border force return to the Pakistan theater precisely because it answered a question the smaller theaters could not. It showed what a powerful state would tolerate, for how long, and at what point the accumulated political cost would finally force a change.

A pattern of the killing across time tells its own story. The first four years were slow, with the annual strike count rarely climbing out of the single digits or low double digits. The years from 2009 to 2012 were the storm, with hundreds of strikes packed into that window and 2010 standing as the single worst year, when the figure crossed a hundred. After 2012 the tempo fell, year over year, a decline from dozens of strikes annually to a couple of dozen, then to single digits, until by 2016 and 2017 the annual figure had dropped to a handful, and by 2018 the campaign had effectively stopped. Plotted on a timeline the program forms a sharp curve, a slow rise, a violent peak, and a long tapering descent. That shape is not random. It reflects, in sequence, the early caution of the Bush years, the deliberate escalation chosen by Obama, and then the combination of factors that brought the program down, which included a thinning of the target pool, rising diplomatic friction, and a shift in American strategy toward Afghanistan and elsewhere.

Geography of the deaths is as concentrated as the chronology. The Federally Administered Tribal Areas, a string of seven agencies running along the Afghan border, absorbed almost the entire campaign, and within that strip the two Waziristans absorbed the overwhelming share. North Waziristan alone accounted for a clear majority of all strikes, and South Waziristan for much of the remainder, leaving only a thin scatter of attacks across the other agencies and an even thinner scatter outside the tribal belt entirely. This concentration had a consequence that the raw numbers can obscure. For the roughly half a million to one million people who lived in the most heavily struck districts, the campaign was not an occasional event in the news. It was a constant overhead presence. Surveys and field interviews conducted by researchers, including the well-known study produced jointly by researchers at Stanford and New York University, described a population that had learned to recognize the sound of a loitering aircraft, that avoided gathering in groups, that hesitated to attend funerals or assist the wounded, and that lived with a chronic, ambient dread. The campaign’s defenders measured it by the leaders it removed. Its critics measured it by the society it frightened. Both measurements were describing the same program.

A particularly grim sub-pattern deserves mention because it shows how the logic of the campaign could turn on itself. Investigations documented strikes that hit the same site twice within a short interval, the second missile arriving after rescuers and neighbors had rushed in to help victims of the first. Reporters and human rights groups called these double-tap strikes, and they also documented attacks that struck funeral gatherings. Whether such strikes deliberately targeted rescuers and mourners or simply struck again on the assumption that militants would be present remained contested. What was not contested was the chilling effect. A population that fears a second strike will not run to pull a wounded child from rubble, and a campaign that produces that hesitation has done something to a society that no single casualty figure can capture.

A problem of the unnamed dead deserves its own mention, because it is the quiet scandal at the center of the statistical record. A substantial share of the people killed in the campaign were never identified by name in any public record. They were counted, in the trackers’ spreadsheets, as a number with a label, so many militants or so many civilians, but the individuals behind the number remained anonymous. The Bureau of Investigative Journalism eventually launched a dedicated effort, sometimes described as a naming-the-dead project, precisely because the anonymity of the casualties was itself a form of erasure. A war fought against named, photographed, dossier-bearing high-value targets produced, as its actual output, thousands of unnamed corpses in remote valleys. That asymmetry, between the precision of the targeting language and the anonymity of most of the dead, is one of the most revealing facts about the campaign, and it is the reason the findable truth of the program lives in a range of numbers rather than in a list of names.

The Decline and the End

The program did not end with a decision or an announcement. It faded. Several forces converged from roughly 2012 onward to pull the campaign down from its peak, and understanding that descent is as important as understanding the escalation.

A first force was attrition of the target pool itself. After years of sustained strikes, the dense roster of senior al-Qaeda and Taliban figures who had clustered in the tribal belt had been thinned. The most valuable targets had been killed, had fled deeper into Afghanistan, or had adopted such severe operational discipline, avoiding phones, avoiding vehicles, avoiding gatherings, that they became extremely difficult to find. A campaign optimized to kill leaders runs short of work when the leaders are gone.

A second force was political, and the year 2011 was the year the political ground gave way. It was a sequence of compounding crises, each one corrosive on its own and devastating in combination. In January a CIA contractor named Raymond Davis shot and killed two men on a Lahore street, was arrested by Pakistani police, and triggered a diplomatic standoff that lasted weeks and ended only with a contested payment to the victims’ families. In May, American special operations forces flew into Abbottabad and killed Osama bin Laden without informing Pakistan in advance, a humiliation for the Pakistani military and a demonstration, separate from the drone campaign, of how thoroughly the United States was willing to violate Pakistani sovereignty when it chose to. Readers tracing how the Abbottabad raid fit into the larger American presence in Pakistan can follow the CIA’s decade-long hunt for bin Laden, which unfolded in parallel with the strike campaign and shared its underlying logic. In November of that same year a NATO air attack near the Salala border post killed twenty-four Pakistani soldiers, an incident that pushed the relationship close to rupture.

Accumulated injuries of 2011 produced concrete retaliation. Pakistan expelled the United States from the Shamsi airbase in Balochistan, a facility that had reportedly served as a launch and recovery site for some of the aircraft, forcing the program to rely entirely on bases outside the country. Pakistan also closed the ground supply routes that fed NATO forces in Afghanistan, a closure that lasted months and cost the alliance dearly before it was reopened. The Pakistani parliament passed resolutions demanding an end to the strikes, and a national consensus hardened around the position that the campaign was an intolerable violation. The strikes continued after 2011, but the political space in which they operated had narrowed sharply, and every subsequent strike now landed in an environment of open hostility rather than quiet, deniable tolerance.

A third force was a shift in American policy and self-presentation. In May 2013 President Obama delivered a major speech at the National Defense University in which he addressed the campaign directly for the first time in such detail, defended its necessity, acknowledged that it had killed people the United States had not intended to kill, and announced tighter internal guidelines meant to restrict strikes to targets that posed a continuing and imminent threat and to require near-certainty that civilians would not be harmed. In the same period the administration acknowledged that four American citizens had died in strikes in the region, including one who had been deliberately targeted in a separate theater. The new guidelines did not end the program, but they signaled that the era of unconstrained escalation was over and that the political cost of the campaign had become heavy enough to require management.

A fourth force was the changing shape of the war itself. As the United States drew down its forces in Afghanistan, the strategic rationale for a high-tempo campaign against the Afghan-focused fighters who used the tribal belt as a rear base weakened. The Pakistani army, meanwhile, finally launched its own large ground operation into North Waziristan in 2014, an offensive that cleared and depopulated much of the territory that had been the program’s center of gravity. With the Pakistani military now operating in force in the very valleys the campaign had struck, the political and practical room for American missiles shrank further. The two trends, the American drawdown and the Pakistani offensive, combined to make the campaign both less necessary and less feasible at the same time.

What the strikes left behind in the tribal belt outlasted the strikes themselves. The 2014 Pakistani army offensive into North Waziristan displaced hundreds of thousands of residents, who became internally displaced persons living in camps and rented rooms far from valleys that had been their families’ home for generations. Some returned years later to find compounds destroyed, orchards untended, and a security architecture of checkpoints and surveillance that had not existed before. The years of overhead aircraft had left a population with measurable psychological scars, documented by the researchers who reached the survivors, including anxiety, sleeplessness, and a learned aversion to gathering in groups that did not simply switch off when the last missile fell. Economic life had been disrupted across a strip of territory that was already among the poorest in Pakistan. Education had suffered, because schools in a strike zone are dangerous places to assemble children. None of this damage appears in a tally of militants removed, and that is precisely the point critics make when they argue that the program’s true cost cannot be read from its target list. A military effort is not only the sum of its strikes. It is also the condition it leaves in the place it was waged, and the condition the strikes left in the tribal agencies was one of displacement, trauma, and a deep, durable grievance against both the foreign power that fired the missiles and the home government that had quietly allowed it. When the 2018 merger folded the agencies into an ordinary province, it changed their legal status but did nothing, on its own, to repair what the preceding fourteen years had done. The legacy of the program is therefore not a closed chapter. It is a set of consequences still working their way through the lives of the people who lived beneath the aircraft.

By the middle of the decade the annual strike count had fallen into the single digits in most years, and the last recorded strikes inside Pakistan came in 2018. That same year Pakistan formally merged the Federally Administered Tribal Areas into the neighboring province of Khyber Pakhtunkhwa through a constitutional amendment, dissolving the special legal status that had governed the tribal agencies for more than a century. The administrative region that had absorbed the entire campaign ceased, on paper, to exist as a distinct entity. The program that had defined a decade of life in those mountains ended not with a treaty or a verdict but with a quiet trailing off, leaving behind a contested body count, a traumatized population, and a precedent that other states were watching closely. For the wider provincial picture into which the tribal agencies were absorbed, the terror landscape of Khyber Pakhtunkhwa shows how the merged districts fit into Pakistan’s larger security map.

Key Figures

A program run from the shadows still had architects, enablers, and victims whose individual stories illuminate the whole. Five figures stand out, each representing a different dimension of how the campaign worked.

Pervez Musharraf

The Pakistani general who seized power in a 1999 coup and ruled until 2008 was the indispensable enabler of the program’s birth. For years Musharraf’s government maintained the public posture that the strikes were an intolerable violation of Pakistani sovereignty, and Pakistani officials condemned them in international forums and to their own citizens. Then, in a 2014 interview with an American magazine conducted after he had left power, Musharraf admitted what had long been suspected. He acknowledged that his government had permitted the Agency to fly armed aircraft over Pakistani territory and to strike, and he described the permission as limited to occasions when a target was isolated and the risk of collateral damage was judged low. His account was self-serving, designed to present the cooperation as careful and constrained, and it almost certainly understated the scope of the arrangement. Yet the admission itself was historically important. It confirmed from the mouth of the head of state that the campaign’s foundational denial had been a deliberate construction, that Pakistan had said no in public while saying yes in private, and that the entire program had been built on a managed lie agreed to by both governments.

George W. Bush

The American president under whom the campaign was born approached it cautiously, as one tool among many in a war he was fighting on several fronts. Bush authorized the first strike in Pakistan in 2004 and roughly fifty over the course of his remaining time in office, a pace that looks restrained only in comparison with what followed. His administration officially denied the extent of the policy, and the program in the Bush years was less a campaign than a series of opportunistic operations against named senior figures when the intelligence and the opportunity happened to align. Bush’s role is easy to underrate because the strike numbers from his presidency are small, but it was under his administration that the foundational arrangement with Pakistan was struck, that the intelligence apparatus was assembled, and that the legal theories justifying killing by remote control were first developed. He built the machine. His successor ran it at full throttle.

Barack Obama

No single individual shaped the campaign more than the American president who inherited it as a minor tool and turned it into a defining instrument of national security policy. Obama’s relationship to the program was complicated and, to his critics on both the left and the right, contradictory. He was a constitutional lawyer who had campaigned against the previous administration’s expansive theories of executive war power, and he became the president who institutionalized targeted killing by remote control on a scale never seen before. He read the intelligence on individual targets, involved himself in the criteria for who could be struck, and defended the campaign as a moral improvement over the alternative of large ground wars. He also presided over the years of heaviest civilian casualties and the expansion of signature strikes. The 2013 National Defense University speech represented his attempt to place the program inside a framework of law and restraint, to make peace with a tool he clearly found troubling but considered necessary. The verdict on Obama and the program depends entirely on the counterfactual the assessor chooses. Measured against doing nothing, the campaign removed dangerous men. Measured against a campaign that killed fewer noncombatants, it fell short. Measured against the large occupying armies it was meant to replace, it was, by the cold calculus of body counts, less lethal.

John O. Brennan

If Obama set the policy, John Brennan operated it. A career intelligence officer who served first as the president’s senior counterterrorism adviser in the White House and then as Director of the CIA, Brennan was the campaign’s chief manager and its most articulate institutional defender. He oversaw the development of the targeting process, the criteria by which names and signatures were nominated for strikes, and the bureaucratic machinery that turned intelligence into a kill decision. Brennan publicly insisted that the program was extraordinarily precise and that civilian casualties were rare, claims that independent trackers and human rights researchers strongly disputed. His role illustrates a structural feature of the campaign worth naming clearly. The program was run primarily by an intelligence agency rather than by the uniformed military, which meant it operated under intelligence rules of secrecy rather than military rules of accountability. That choice of institution helped keep the campaign deniable, kept its records classified, and kept it insulated from the kind of after-action review that governs conventional military operations. Brennan was the human embodiment of that choice.

Baitullah Mehsud

Among the thousands killed, Baitullah Mehsud stands as the figure whose death best demonstrates what the program could achieve when intelligence, opportunity, and shared interest aligned. The founder of the Pakistani Taliban was not primarily a threat to the United States. He was a threat to Pakistan, the organizer of a campaign of suicide bombings that killed Pakistani soldiers, police, and civilians, and the man many held responsible for the assassination of Benazir Bhutto. When a missile killed him in South Waziristan in August 2009, the strike removed a genuine architect of mass violence and did so in a way that served Islamabad’s interests directly. His killing is the strongest single piece of evidence the program’s defenders can point to, a clean removal of a clearly dangerous man. It also raises the sharpest version of the program’s central question. If the campaign worked well enough to kill a figure like Baitullah, why did the violence in the tribal belt not subside, and why did the Pakistani Taliban survive his death, reconstitute under new leadership, and continue killing for years afterward. The answer to that question is the analytical heart of the entire debate.

Consequences and Impact: Sovereignty, Consent, and Backlash

The program’s most lasting consequences were not measured in militants removed. They were measured in what the campaign did to law, to the relationship between two governments, and to the population that lived beneath the aircraft.

Begin with the paradox at the program’s core, the gap between Pakistan’s public position and its private conduct. For fourteen years Pakistan’s official line was consistent and emphatic. The strikes violated Pakistan’s sovereignty, breached international law, inflamed its population, and must stop. Pakistani diplomats raised the issue at the United Nations. Pakistani parliamentarians passed resolutions. Pakistani crowds burned American flags. And throughout, as Musharraf’s later admission confirmed and as a substantial body of leaked cables and reporting indicated, elements of the Pakistani state were sharing intelligence that helped guide the strikes, were permitting the overflights, and in the early years were hosting the aircraft on Pakistani bases. The consent was real but it was also genuinely limited, contested within the Pakistani state, and subject to change. It is most accurate to say that Pakistan’s cooperation was partial, deniable, internally divided, and conditional, and that the public protest was not entirely theater. A government can simultaneously permit something it cannot afford to be seen permitting. That is precisely the position Pakistan occupied, and the strain of holding that position for fourteen years corroded the credibility of the Pakistani state with its own citizens.

Why would any government accept such a corrosive arrangement? The answer reveals how the consent worked in practice. Pakistan’s military gained several things from the campaign that it could not easily obtain otherwise. It got the elimination of figures, such as Baitullah Mehsud, who were waging war on the Pakistani state itself, and it got those eliminations without having to send its own soldiers into the most dangerous valleys to do the job. It got intelligence and equipment from the United States as part of the broader relationship. And it got to direct, at least sometimes, where the missiles fell, which meant the campaign could be steered toward the army’s enemies and away from the militant groups the army still regarded as useful. The arrangement was humiliating in public and useful in private, and a security establishment that valued the usefulness more than it feared the humiliation kept the bargain alive for more than a decade.

A sovereignty question the program raised was never resolved, only exploited. Under the United Nations Charter the territory of a state is not to be attacked without that state’s consent or without the authorization of the Security Council, with a contested exception for self-defense. The United States justified the campaign primarily on a theory of self-defense against al-Qaeda and associated forces, combined with the argument that Pakistan was unwilling or unable to address the threat on its own territory, an argument that becomes legally awkward when the same unwilling or unable host is quietly cooperating. The campaign therefore lived inside a legal ambiguity that no court ever cleanly adjudicated. The deeper analysis of that ambiguity, and of how different states have constructed competing justifications for killing abroad, belongs to the global legal debate over targeted killing, but the relevant point here is simpler. The program did not resolve the law. It demonstrated that a sufficiently powerful state could operate in the gaps of the law for more than a decade and absorb the criticism.

Backlash inside Pakistan was severe and is the strongest argument the program’s critics deploy. The strikes were ferociously unpopular across the Pakistani political spectrum. They handed militant recruiters a propaganda gift, the image of an arrogant foreign power killing Pashtuns from the sky with impunity. They fed a narrative, widespread in Pakistan, that the country’s sovereignty was worthless and its government a junior partner in its own humiliation. Anti-Americanism in Pakistan, already high, hardened into something close to a national consensus. The campaign also became entangled with the rise of the Pakistani Taliban’s war against the Pakistani state, and while it would be too simple to say the strikes created that insurgency, they were one ingredient in a toxic mixture of grievance, ideology, and state failure. The strategic ledger here is genuinely difficult. The program removed dangerous individuals, and it also generated political consequences that arguably made the broader struggle harder. Reasonable analysts have weighed those two facts and reached opposite conclusions.

There was also a consequence for the wider world, and it is the one most relevant to everything that came after. The campaign normalized a practice. Before the Pakistan program, the sustained, systematic killing of named individuals across an international border by a state that refused to confirm it was an exotic act. After fourteen years of the Pakistan campaign, it was an established instrument of statecraft, studied in war colleges, copied in technique, and absorbed into the strategic vocabulary of every government with the means and the motive to consider it. The United States did not invent extraterritorial killing. But it industrialized it, gave it a technology, and proved its political survivability. Other states took notes.

Normalization had a hard, physical dimension as well as a political one, and it is visible in the spread of the technology itself. When the Pakistan campaign began, armed remotely piloted aircraft were effectively a monopoly of a single power. By the time the campaign wound down, that monopoly had collapsed. Other states had acquired, built, or imported armed aircraft of their own, and a brisk international market in the technology had emerged, with several manufacturing nations willing to sell systems that the original developer had been reluctant to export. The doctrine spread alongside the hardware. A generation of military and intelligence officers around the world had watched the American program, studied its tactics, absorbed its vocabulary of targeting and signatures and persistence, and concluded that killing across a border from the air was a feasible and survivable instrument of state. That diffusion is one of the program’s most consequential legacies, and it is largely irreversible. A technology that has been demonstrated, productized, and sold cannot be uninvented, and a doctrine that has been shown to work cannot be unlearned. The United States, in conducting the most visible sustained campaign of its kind, did not only fight its own war. It also ran a fourteen-year demonstration, witnessed by every government with an interest in the subject, of what the technology could do and what political weather it would have to endure. The lesson other states drew was not uniform. Some concluded that the visible, attributable method was worth the diplomatic cost. Others, watching the same demonstration, concluded the opposite, that the goal was sound but the loud instrument was not, and that a quieter tool would serve them better.

A campaign of this kind also reshaped politics inside Pakistan in ways that outlasted the strikes. The issue of the American aircraft became a fixture of Pakistani electoral and parliamentary debate, a reliable subject for opposition politicians who could attack the sitting government for complicity and weakness. A political movement built part of its national profile on loud opposition to the strikes, organizing rallies and road blockades and turning the issue into a symbol of everything that was wrong with Pakistan’s relationship with Washington. The military, which had quietly enabled the program, found itself publicly criticized for failing to defend the country’s airspace, an awkward position for an institution that grounded its prestige on being the guarantor of national security. Civilian governments were squeezed between an American partner they could not afford to alienate and a public that regarded the strikes as a national humiliation. The strikes thus became a permanent irritant in the country’s internal politics, a wedge that opponents could drive between the government and the governed, and a standing reminder that Pakistan’s sovereignty had limits its own citizens found intolerable. None of this serves a simple verdict. A program intended as a narrow counterterrorism tool ended up as a significant variable in the political life of a nation of two hundred million people, feeding cynicism about the state, fueling anti-American sentiment that politicians could harvest, and corroding trust between Pakistanis and the institutions that govern them. That is a large and lasting consequence for an effort whose official purpose was simply to kill a roster of men in a few mountain valleys, and it belongs in any honest accounting of what the program actually did.

The Effectiveness Debate: Degradation Versus Radicalization

The hardest question about the program is also the most important, and it is the question the deep analysts of counterterrorism have argued for fifteen years without consensus. Did the campaign work? The dispute is best understood as a contest between two genuine bodies of evidence, one supporting the conclusion that the strikes meaningfully degraded terrorist organizations, the other supporting the conclusion that the strikes created more enemies than they removed.

Take the degradation case first, because it has real force. By the metric of leadership attrition the program was undeniably productive. Over fourteen years it killed a long roster of senior al-Qaeda figures, including operational planners and the men who managed the organization’s external operations. It killed successive leaders of the Pakistani Taliban, Baitullah Mehsud in 2009 and Hakimullah Mehsud in 2013, and it kept killing replacement commanders fast enough that the organizations struggled to maintain continuity at the top. Analysts who study the degradation question, including scholars associated with Brookings who have examined whether decapitation strikes weaken terrorist groups, have argued that the cumulative pressure imposed real costs. It forced militant leaders into hiding, disrupted their communications, made it dangerous for them to train, plan, and gather, and consumed enormous amounts of their time and attention simply on survival. Al-Qaeda’s central organization, the entity that planned the September attacks, was a hollow shadow of its former self by the middle of the decade, and the campaign was one of the instruments that hollowed it. By the standard of preventing another mass-casualty attack on the American homeland planned from the tribal belt, the program’s defenders can point to a record of success that is hard to dismiss.

A subtler version of the degradation argument is worth stating, because it does not depend on body counts at all. Even when a strike missed, even when it killed a replaceable foot soldier rather than an irreplaceable leader, the mere fact of a constant overhead threat imposed what analysts call a tax on the enemy’s operations. A militant organization that must assume it is being watched cannot run large training camps in the open, cannot convene big planning meetings, cannot use phones freely, and cannot move its leaders around without elaborate precautions. Every one of those constraints slows the organization down and makes it less capable of planning the complex, large-scale operations that pose the gravest danger. On this view the campaign’s value lay less in the specific men it killed than in the operating environment it created, an environment of permanent insecurity that made sophisticated terrorism harder to organize. It is a real argument and a serious one.

Now take the radicalization case, which has equal force and a different shape. Critics, including the counterinsurgency theorist David Kilcullen, whose work on how outside intervention generates what he called the accidental guerrilla has shaped a generation of thinking, argued that the strikes were strategically self-defeating in the longer run. The argument runs as follows. Each strike that killed a civilian, or that the local population believed had killed a civilian, generated grief, rage, and a powerful motive for revenge among the dead person’s family and clan. In a tribal society organized around honor and kinship, killing a man’s brother or son creates an obligation of retaliation that does not expire. The campaign therefore functioned, on this view, as a recruitment engine for the very organizations it was meant to destroy, trading the removal of one fighter today for the radicalization of several tomorrow. The signature strikes were especially corrosive here, because killing people whose names were unknown guaranteed, statistically, that some of the dead would be noncombatants, and each such death rippled outward through a network of relatives. The critics also pointed to the simple fact that after fourteen years the militant ecosystem of the region was still functioning, still violent, and still capable of reconstituting its leadership, which suggested that the strategy, whatever its tactical successes, had not achieved a strategic defeat of the enemy.

There is also a moral dimension to the radicalization case that the strategic language can flatten. A campaign that kills children, that frightens an entire population away from funerals and rescue work, and that offers the bereaved no acknowledgment and no recourse does damage that is not captured in any ledger of organizational degradation. Even an analyst persuaded that the strikes degraded al-Qaeda must reckon with the cost paid by the people of the tribal belt, who did not choose to host the militants, did not vote on the campaign, and could not appeal it. The effectiveness debate is usually conducted in the cold vocabulary of strategy, but the program’s critics insist, correctly, that strategy is not the only frame, and that a method which works against an organization can still be wrong against a population.

Underneath the whole effectiveness dispute lies a measurement problem that neither side can fully escape. To know whether the strikes worked, an analyst would need to compare what actually happened with what would have happened had the campaign never been waged, and that counterfactual world is unobservable. A defender can point to the absence of a successful mass-casualty attack on the American homeland planned from the tribal belt and call it proof of success, but the absence of an event is weak evidence, because the event might not have occurred even without the strikes. A critic can point to the militant ecosystem that survived fourteen years of bombardment and call it proof of failure, but survival under pressure is not the same as flourishing, and the organization that endured was, by most accounts, a weaker thing than the organization that had existed before. Each side, in other words, is reading the same ambiguous record through the lens of its prior conviction. This is not a reason to throw up one’s hands. It is a reason to be precise about what can and cannot be claimed. What can be claimed with confidence is narrow and concrete. The strikes killed specific, identifiable senior figures. They imposed real operational costs on the organizations that lost those figures. They also killed noncombatants, frightened a population, and generated lasting grievance. What cannot be claimed with confidence is the grand verdict, the single sentence declaring the program a success or a failure, because that verdict depends on a counterfactual no one can observe. The most honest posture is to hold the concrete findings firmly and the grand verdict loosely, and to be suspicious of anyone, defender or critic, who offers the grand verdict with too much certainty.

How should the dispute be adjudicated? The honest answer is that both bodies of evidence are partly correct and that the campaign’s success depends on which enemy and which threat one chooses to measure. Against the specific threat of a spectacular al-Qaeda attack on the United States planned and directed from a Waziristan sanctuary, the program plausibly worked. That sanctuary became a far harder place to run a global terrorist organization, and the central organization decayed. Against the broader objective of pacifying the region, defeating militancy as a phenomenon, and leaving the tribal belt more stable than it found it, the program plainly failed, and may even have been counterproductive. The campaign was a good tool for the narrow job of disrupting a particular organization and a poor tool for the wide job of winning a war. Its defenders and its critics have been arguing past each other for years partly because they are scoring two different games. That distinction, between degrading an organization and defeating an insurgency, is the single most useful idea to carry away from the effectiveness debate, and it is directly relevant to the comparison that follows.

Why It Still Matters: The Method India Did Not Choose

The most important reason to study the American campaign in Pakistan today is that it became the reference case, the worked example, against which every subsequent program of cross-border killing has been measured. When India’s covert campaign against its own most wanted began to draw attention, the natural comparison was to the American program, because both states were doing something structurally similar, killing terrorists on Pakistani soil. Yet the two campaigns chose opposite instruments, and the contrast between them is one of the most instructive in modern security policy.

The American method was the missile fired from a remotely piloted aircraft, controlled from a base in Nevada, watching its target through a camera for hours before striking. The method India is widely reported to use is the close-range shooting, two men on a motorcycle, a pistol or a short burst at point-blank distance outside a mosque or on a morning walk, followed by an escape through congested streets. The full anatomy of that signature is examined in the decoded pattern of the unknown gunmen, and the systematic comparison of the two national approaches is the subject of the dedicated study of the US drone campaign set against India’s shadow war. The point worth making here is why the difference exists, and the answer is that India studied the American program and learned from its costs.

Consider what the American method required that the Indian method does not. The campaign needed airspace access, which meant it needed at least the tacit permission of the host government, permission that could be and eventually was withdrawn. It needed launch and recovery bases, including, for years, a base on Pakistani soil from which the United States was ultimately expelled. It needed satellite communications, a chain of bases, a fleet of expensive aircraft, and an enormous apparatus of analysts and operators. Above all, it could not be denied. A missile strike that destroys a compound and kills a dozen people is a loud, visible, attributable act of war. The whole world knew the United States was conducting the campaign, and that visibility was the source of the program’s gravest costs, the diplomatic crises, the sovereignty confrontations, the propaganda victories handed to the enemy, the corrosion of the host government’s standing with its own people.

India, examining that ledger, appears to have concluded that the costs of the visible method were not worth paying and that the same strategic objective could be pursued with an instrument that generated almost none of them. A motorcycle and a pistol require no airspace, no base, no satellite, no fleet, no permission. A close-range shooting in a crowded city can be attributed to a criminal feud, a sectarian dispute, or an internal militant rivalry, and the killing state can shrug and deny with a straight face, because there is no missile crater, no aircraft wreckage, no electronic signature to contradict the denial. Where the American program optimized for reach and tempo, the ability to kill many targets per month across a wide area, the Indian method optimizes for deniability and political survivability, accepting a far slower pace in exchange for the near-total absence of diplomatic blowback. India’s wider doctrine, the logic of choosing the quiet instrument over the loud one, is laid out in the overview of India’s shadow war against terror, and the pattern of specific eliminations, including operations carried out as far afield as the tribal districts where a Masood Azhar aide was shot dead in North Waziristan, shows the method in repeated practice.

There is a deeper trade hidden inside the comparison, and it is worth naming directly. The American method bought scale at the price of attribution. A drone fleet can strike many targets a month across a wide theater, but it cannot pretend it was not there. The Indian method buys deniability at the price of scale. A pair of gunmen can kill one carefully chosen target at a time, but it can be disowned completely. Each state, in effect, optimized for the variable it valued most. The United States, fighting what it regarded as an active threat to its homeland, valued tempo and reach and was willing to pay the diplomatic bill. India, pursuing a long campaign against a neighbor with whom it must coexist indefinitely, valued the ability to deny and was willing to accept a slower pace. The two campaigns are not better and worse versions of the same thing. They are two different answers to the question of what a state is willing to sacrifice in order to kill its enemies abroad.

There is one more reason the Pakistan precedent reaches into the present, and it concerns the question of what a watching state could not learn from the American example. The campaign demonstrated that cross-border killing was possible and survivable, and it demonstrated which features of the loud method generated the heaviest costs. What it could not demonstrate was whether a quieter method would actually work as well, because no quiet method had yet been run at comparable scale and scrutiny. A state choosing the motorcycle over the missile was therefore making an informed bet rather than following a proven recipe. It was wagering that the deniable instrument could degrade an enemy organization with something close to the effectiveness of the visible one, while shedding the diplomatic, legal, and reputational bills that the visible one had run up. The American program supplied half the lesson, the half about costs, with unusual clarity. The other half, about whether the quiet alternative could match the loud one’s strategic output, remained an open question that only a later campaign could answer. That is why the American experience and the campaigns that followed it are best read together rather than apart. The first established the possibility and exposed the price. The second is, in effect, the test of whether the price was avoidable. Studied as a pair, the two illuminate the central choice facing any state that decides, rightly or wrongly, to kill its enemies on foreign soil, the choice between reach and deniability, between tempo and quiet, between a method that cannot be hidden and a method that can be disowned. The Pakistan campaign did not settle that choice. It defined the terms, and it left the terms on the table for every government that came afterward to weigh for itself.

The contrast carries a final lesson that reaches back to the effectiveness debate. The American campaign was loud, attributable, and infrastructure-heavy, and it therefore paid a heavy strategic price in legitimacy even when it succeeded tactically. The lesson a watching state could draw was not that cross-border killing does not work, but that cross-border killing works best when it cannot be proven. The United States demonstrated, across fourteen years, that the territory of a nuclear-armed neighbor could be struck repeatedly without catastrophe. It also demonstrated, by the size of the bill it paid, exactly which features of its method generated that bill. A later actor, free to design its own campaign with the American example in front of it, could keep the demonstrated possibility and discard the demonstrated costs. That is, in the end, why the American program in Pakistan still matters. It was not only a campaign. It was a fourteen-year experiment whose results other states have read carefully, and the quiet killings that continue on Pakistani soil today are, in part, the experiment’s conclusion put into practice.

Frequently Asked Questions

Q: How many US drone strikes hit Pakistan?

The most widely cited dataset, maintained by the Bureau of Investigative Journalism, recorded at least four hundred and thirty confirmed strikes inside Pakistan between the first attack in June 2004 and the last recorded strikes in 2018. Other trackers, including the Long War Journal and the New America Foundation, produced figures in a broadly similar range. No exact official count exists because the Central Intelligence Agency never published a strike log and Pakistan never released a casualty record for its own tribal districts. The four hundred and thirty figure should therefore be understood as a careful minimum compiled from open sources rather than a precise government total. The great majority of those strikes were concentrated in a narrow band of years from 2008 to 2012, with the single year 2010 standing as the most intense twelve-month period of the entire campaign.

Q: How many people were killed by US drones in Pakistan?

Estimates of total deaths fall in a range of roughly two thousand five hundred to four thousand. The Bureau of Investigative Journalism, whose figures are the most frequently cited, recorded a span from about two thousand five hundred to slightly above four thousand. The width of that range is not a flaw in the research. It reflects the genuine difficulty of counting the dead in a closed war zone where journalists were largely barred, where competing official and local accounts could not be reconciled, and where many of the dead were never publicly named. The range itself is the most honest single answer. The campaign ended thousands of lives, and the exact figure will never be fixed because the program was never designed to permit a verifiable accounting.

Q: Did Pakistan secretly agree to the drone program?

Yes, though the agreement was partial, internally divided, and deliberately deniable. For fourteen years Pakistan’s official position was that the strikes violated its sovereignty and must stop, and Pakistani officials condemned the campaign publicly. Behind that public posture, elements of the Pakistani state shared intelligence that helped guide strikes, permitted the overflights, and in the early years hosted American aircraft on Pakistani bases. The clearest confirmation came from former president Pervez Musharraf, who admitted in a 2014 interview that his government had permitted the Agency to fly armed aircraft over Pakistani territory. The most accurate description is that Pakistan said no in public while saying a limited and conditional yes in private, and that the cooperation was contested within the Pakistani government itself.

Q: How many civilians were killed by US drones in Pakistan?

Civilian death estimates vary dramatically by source, which is itself an important fact about the campaign. The Bureau of Investigative Journalism estimated civilian deaths in a range from several hundred to nearly a thousand, and estimated that somewhere between one hundred seventy and two hundred of the dead were children. The Long War Journal, using a narrower definition of who counted as a noncombatant, recorded substantially lower civilian tolls. The New America Foundation produced intermediate figures. The divergence stems partly from definitional disputes, because the United States at times treated military-age men in a strike zone as presumptive combatants unless proven otherwise. No single number can be stated with confidence. What is clear is that a meaningful fraction of the dead were not fighters, and that the campaign killed children.

Q: What was the first US drone strike in Pakistan?

The opening attack took place on June 18, 2004, near Wana in South Waziristan. It killed a tribal militant commander named Nek Muhammad Wazir along with several others. Nek Muhammad had fought the Pakistani army to a standstill and had become an embarrassment to the military, so his death served Islamabad’s interests as well as Washington’s. In a detail that set the template for the entire program, Pakistan’s army initially claimed the strike as its own operation, a fiction that established the pattern of the foreign power doing the killing while the host government absorbed or deflected the attribution. Every strike for the next fourteen years operated within the fog of official denial that this first attack created.

Q: Where in Pakistan did most drone strikes take place?

The overwhelming majority of strikes landed in the Federally Administered Tribal Areas, a string of seven agencies running along the Afghan border. Within that strip, North Waziristan and South Waziristan absorbed the largest share by a wide margin. North Waziristan in particular, and the area around the town of Miranshah, appears repeatedly in the strike record because it had become the densest sanctuary for foreign al-Qaeda fighters, the Haqqani network, and factions of the Pakistani Taliban. This extreme geographic concentration meant that for the population of a handful of districts the campaign was a constant overhead presence, while for most of Pakistan it was a distant news event.

Q: What is a signature strike?

A signature strike is an attack carried out without the operators knowing the identity of the people they are about to kill. Instead of targeting a specific named individual whose identity has been confirmed by intelligence, a signature strike targets people who exhibit a pattern of behavior, a signature, that analysts associate with militancy, such as gathering at a particular kind of compound, handling what appears to be weapons, or traveling a particular route. The United States expanded the use of signature strikes substantially after 2008, and that expansion is the single most important reason the campaign’s tempo rose so steeply. It is also the single most important reason civilian casualty figures became so contested, because when you kill people whose names you do not know, you cannot be certain afterward exactly who they were.

Q: Why did the United States use drones instead of ground operations?

Several reasons converged. A ground operation by American forces inside Pakistan would have meant invading the territory of a nominal ally, an act with enormous diplomatic and military risk, as the lone exception of the Abbottabad raid demonstrated. A large Pakistani military operation in the tribal belt was something Islamabad was reluctant to mount because it would be bloody, unpopular, and politically dangerous. A remotely piloted aircraft solved both problems. It could loiter over a target for many hours, strike with a precision-guided missile, and put no American personnel at physical risk. For an administration that wanted to fight al-Qaeda without the cost of large occupying armies, the technology offered what looked like a cleaner, lighter, less expensive instrument.

Q: What aircraft did the CIA use for the strikes?

The campaign relied primarily on the MQ-1 Predator and, later, the larger and more capable MQ-9 Reaper, both produced by General Atomics. These are long-endurance remotely piloted aircraft, machines without an onboard pilot, flown by operators at consoles thousands of miles away, with a flight crew at one location and intelligence analysts at others. The standard weapon was the laser-guided Hellfire missile, a munition originally designed to be fired from helicopters against armored vehicles. The aircraft were operated under the operational control of the Central Intelligence Agency rather than the uniformed military, a choice that helped keep the campaign classified, deniable, and insulated from conventional military accountability.

Q: Did the United States ever publicly admit to the drone campaign?

For most of the program’s life the United States neither confirmed nor denied the strikes, treating the entire campaign as a classified covert action. The George W. Bush administration officially denied the extent of the policy. The shift came under President Barack Obama. In a major speech at the National Defense University in May 2013, Obama addressed the campaign directly and at length for the first time, defended its necessity, acknowledged that it had killed people the United States had not intended to kill, and announced tighter internal guidelines. In the same period the administration acknowledged that four American citizens had died in strikes in the region. Even after that speech, however, the program retained much of its official secrecy, and a full, verifiable accounting was never released.

Q: Did the drone program actually degrade al-Qaeda?

By the measure of leadership attrition, the campaign was genuinely productive. It killed a long roster of senior al-Qaeda figures, including operational planners and managers of external operations, and the sustained pressure forced surviving leaders into hiding, disrupted their communications, and made it dangerous for them to train, plan, and gather. By the middle of the decade al-Qaeda’s central organization, the entity that planned the September 2001 attacks, was a hollow shadow of its former self, and the campaign was one of the instruments that hollowed it. Critics counter that the strikes also generated grievance and revenge that aided recruitment, and that the broader militant ecosystem survived. The fairest verdict is that the program was effective at the narrow task of degrading a particular organization and far less effective at the wide task of defeating militancy as a regional phenomenon.

Q: What is the TTP and was it targeted by drones?

The Tehrik-i-Taliban Pakistan, commonly called the Pakistani Taliban or the TTP, is an umbrella movement of militant factions waging war against the Pakistani state itself, distinct from the Afghan Taliban. It was responsible for a long campaign of suicide bombings and attacks inside Pakistan. The TTP was indeed targeted by the American campaign, and two of its leaders were killed in strikes, founder Baitullah Mehsud in August 2009 and his successor Hakimullah Mehsud in November 2013. Because the TTP was an enemy of Islamabad as well as a target of Washington, strikes against it represented the rare occasions when American and Pakistani interests aligned almost perfectly, and Pakistan’s objections to those particular strikes were noticeably muted.

Q: How did the drone strikes affect ordinary life in the tribal areas?

For the population of the most heavily struck districts, perhaps half a million to a million people, the campaign was not an occasional event but a constant ambient condition. Field research, including a well-known study by researchers at Stanford and New York University, described a society that had learned to recognize the sound of a loitering aircraft, that avoided gathering in groups, that hesitated to attend funerals or rush to help the wounded for fear of a follow-up strike, and that lived with chronic dread. Documented strikes that hit the same site twice within a short interval, after rescuers had arrived, deepened that fear. The psychological and social toll on the civilian population is one of the campaign’s most serious and least quantifiable consequences.

Q: Why did the drone program slow down after 2014?

Several forces converged to pull the campaign down from its peak. The pool of high-value targets had been thinned after years of sustained strikes, so a campaign optimized to kill leaders ran short of work. The political relationship between Washington and Islamabad deteriorated sharply, especially during the crises of 2011, which included the Raymond Davis affair, the Abbottabad raid, and the Salala border incident, after which Pakistan expelled the United States from a base it had used. The Pakistani army launched its own large ground offensive into North Waziristan in 2014, clearing much of the program’s core territory. American policy itself shifted, with the 2013 National Defense University speech signaling new internal restrictions. The combination of fewer targets, worse diplomacy, a Pakistani offensive, and tighter rules produced a steady year-over-year decline until the last recorded strikes in 2018.

Q: Did the drone program violate international law?

The legality was contested throughout and was never cleanly adjudicated by any court. Under the United Nations Charter, the territory of a state generally may not be attacked without that state’s consent or Security Council authorization, with a disputed exception for self-defense. The United States justified the campaign on a theory of self-defense against al-Qaeda and associated forces, combined with the argument that Pakistan was unwilling or unable to address the threat itself. Critics argued that the campaign violated Pakistani sovereignty and that signature strikes in particular could not satisfy the legal requirements of distinction and proportionality. The campaign lived inside a genuine legal ambiguity that a sufficiently powerful state was able to exploit for more than a decade.

Q: What happened to the tribal areas after the strikes ended?

In 2018, the same year the campaign effectively stopped, Pakistan formally merged the Federally Administered Tribal Areas into the neighboring province of Khyber Pakhtunkhwa through a constitutional amendment. This dissolved the special legal status that had governed the tribal agencies for more than a century and folded them into Pakistan’s ordinary provincial administration. The region that had absorbed nearly the entire campaign therefore ceased, on paper, to exist as a distinct entity. The merger did not erase the campaign’s legacy, a contested death toll, a traumatized population, and damaged infrastructure, but it did mark the formal end of the unusual administrative arrangement that had made the tribal belt so distinctive and so vulnerable.

Q: Why were so many of the dead never identified by name?

A large share of those killed in the campaign were never publicly identified, counted in the trackers’ datasets only as anonymous figures under the labels militant or civilian. The reasons were structural. The strikes hit remote valleys that journalists could not safely reach, the dead were often buried quickly according to local custom, and neither the United States nor Pakistan released identifying records. The Bureau of Investigative Journalism eventually launched a dedicated effort to name the dead precisely because their anonymity was itself a form of erasure. The contrast is striking. A campaign that justified itself with named, dossiered high-value targets produced, as its actual output, thousands of unnamed corpses, and that gap between the precision of the language and the anonymity of the dead is one of the program’s most revealing features.

Q: How does the US drone program compare to India’s shadow war in Pakistan?

Both involved a state killing terrorists on Pakistani soil, but the two campaigns chose opposite instruments. The American method was the missile fired from a remotely piloted aircraft, an act that was loud, attributable, and dependent on visible infrastructure, including airspace access, bases, satellites, and an expensive fleet. That visibility generated heavy diplomatic costs. The method India is widely reported to use, close-range shootings by men on motorcycles, requires almost no infrastructure, can be plausibly attributed to a criminal or sectarian dispute, and is therefore far more deniable. The American program optimized for reach and tempo at the price of attribution. The Indian method optimizes for deniability and political survivability at the price of speed. India appears to have studied the American campaign and deliberately chosen the quieter instrument precisely to avoid the costs the visible one imposed.