When an American soldier dies in Afghanistan, his death is not anonymous. The tragedy of that loss is mourned, and his life is remembered and celebrated. In many cases, the death is covered prominently in local and state media, often for several days. The Pentagon dutifully records the loss, medals are delivered, a ceremonial flag is presented to survivors, and the Defense Department pays the soldier’s family $100,000 in compensation, plus back pay, insurance, housing allowances and more.

But when an Afghan dies in the war — especially an Afghan civilian — her death is rarely noticed by the outside world. Often, it’s not even recorded by Afghan hospitals or morgues. Asked whether his country keeps records of civilian casualties, Said Jawad, the former Afghan ambassador to the United States, sighs. “In Afghanistan, you know, we don’t even have birth certificates,” he says. “Do you know we don’t even have a list of Afghan soldiers and police, members of the security forces, who are killed?”

Most Americans strongly supported the 2001 invasion of Afghanistan, but they have long since turned sharply against a conflict that, in September, entered its thirteenth year — by some measures the longest war in American history. A big reason for the shift in public opinion is the steadily growing list of dead and maimed soldiers, sailors, airmen and marines. For politicians, it’s by now rote to declare that the war has cost the United States more than 2,200 dead, thousands more wounded and at least $640 billion. But even among staunchly antiwar politicians and pundits, few bother to mention the cost to Afghans. “It’s just not part of American discourse,” says John Tirman, author of The Deaths of Others: The Fate of Civilians in America’s Wars. “You don’t have politicians standing up for civilians.”

It is to correct this unconscionable oversight that The Nation has prepared this report. In this special issue, we focus primarily on those who have died at the hands of the United States and its allies. That’s because Americans, collectively, should be accountable for the violence committed in their name. We should demand that our military act humanely and with a determination to avoid civilian casualties.

A large number of these civilian deaths — perhaps most — have come at the hands of the Taliban and its allies. Since gathering momentum in 2006, the insurgents have shown a reckless disregard for civilians, planting tens of thousands of improvised explosive devices along roadways, setting off suicide bombs in crowded marketplaces, and assassinating countless local officials, tribal leaders and other civilians. As the war evolved, civilian casualties attributable to the insurgents rose from a relatively small number in the years before the insurgency really got under way in the mid-2000s, to 55 percent of civilian deaths, according to the 2008 report of the United Nations Assistance Mission in Afghanistan (UNAMA), to what the UN now says is approximately 80 percent of all civilians killed.

As we shall see, even coming up with rough estimates of civilian casualties is difficult. But it’s an unassailable fact that many of those killed by anti-government forces would almost certainly be alive had the United States never invaded. And the victims of US forces and other foreign troops number in the many thousands. The United States has been singularly uninterested in tracking and accounting for the Afghan dead, whether civilians or combatants. In an echo of the discredited metrics of the Vietnam War era, Gen. Tommy Franks, who led the US invasion in 2001 and served as commander of Central Command (CENTCOM) from 2000 to 2003, was even more blunt. “You know we don’t do body counts,” he said.

In 2008, the International Security Assistance Force (ISAF), the US-led coalition in Afghanistan, reluctantly began to track civilian casualties, setting up the Civilian Casualty Tracking Cell and other mechanisms to do so. But as we report below, this work was woefully incomplete. At the same time, advocates of the vaunted counterinsurgency doctrine promoted by Gen. Stanley McChrystal and then, less stringently, by Gen. David Petraeus, along with academics such as Sarah Sewall of Harvard, instituted a new policy that emphasized the protection of civilians. But despite the policy changes, Afghan civilians continued to die in large numbers — a situation acknowledged by Tim Rieser, a top aide to Senator Patrick Leahy. “No matter how much the Pentagon says that they’re going to revise their tactics or their procedures or whatever, [the deaths] keep happening!” says Rieser, whose boss has been a leading voice for decades on human rights.

I. Counting the Dead

Iraq, which endured an eight-year war with Iran, followed by the Gulf War, a dozen years of lethal US-engineered sanctions, the 2003 US invasion and a civil war, still maintains a functioning system of hospitals, clinics and morgues, and researchers are able to make use of roughly accurate demographic data based on household surveys. One such study, published in The Lancet in 2006, estimated, not without controversy, more than 600,000 “excess deaths” resulting from the US war and occupation. There is no parallel study for Afghanistan, according to Neta Crawford, a political scientist at Boston University who has written extensively on civilian deaths in Afghanistan and who has tried to raise funds to conduct a household survey there.

The Asia Foundation, which conducts an annual Survey of the Afghan People, has perhaps come closest to gauging the war’s toll. Based on more than 6,300 interviews with adult Afghans in all thirty-four provinces, the survey reports that over one-fifth (22 percent) of the population — more than 6 million people — personally experienced some kind of crime or violence in their household in 2011. Of those, 8 percent (about 500,000 people) report having suffered violence at the hands of “foreign forces” — i.e., ISAF. And those figures are just a one-year snapshot. Multiply that by twelve years of war, and it becomes evident that millions of Afghans have suffered death, injury, and damage to their homes or livelihoods by US and ISAF forces.

The United Nations, which began to track civilian casualties systematically by 2008, around the same time as the US military and ISAF, arguably did a somewhat better job than the latter — but former UN officials interviewed by The Nation say that even the UN, with trained investigators and many offices spread across the country, managed to track only a portion of those killed. A handful of underfunded local NGOs, including the Afghanistan Independent Human Rights Commission and the Afghan NGO Security Office, have monitored the conflict, but they’ve failed to produce reliable counts. And the Afghan government hasn’t been able to keep track of the war’s human cost.

NGOs outside Afghanistan, including Human Rights Watch, the Center for Civilians in Conflict and the Open Society Foundations, have made valiant efforts to track and document abuses, human rights violations, war crimes and major mass-casualty incidents, but none have maintained a database of civilian or combatant deaths (the Bureau of Investigative Journalism, based in Britain, has compiled extensive data on civilian casualties worldwide resulting from US drone strikes, but not overall civilian casualties in the Afghan War). For some time, Professor Marc Herold of the University of New Hampshire doggedly tracked Afghan civilian casualty incidents, but he included in his data what most analysts say are exaggerated or fabricated reports from the often pro-Taliban Pakistani media. So far, perhaps the best account of casualties was part of the “Costs of War” report prepared by Brown University’s Watson Institute for International Studies, under the direction of Catherine Lutz and Neta Crawford. Crawford’s paper “Civilian Death and Injury in Afghanistan, 2001–2011,” updated in February 2013, estimates as many as 19,000 civilians killed by all sides, and she provides a valuable compendium of estimates for combatant deaths too. Still, Crawford’s estimate of civilian casualties relies heavily on the reports of UNAMA, which understates the total number of deaths significantly.

A central part of The Nation’s project on civilian casualties in Afghanistan has been to compile a uniquely comprehensive interactive database of civilian casualty incidents from the beginning of the war in 2001 to the end of 2012. It includes information gleaned from reliable media accounts — in outlets such as the New York Times, the Guardian and CNN — of 458 separate incidents, involving between 2,848 and 6,481 people, who died as a result of war-related actions by the United States, its allies and Afghan government forces. It includes high-profile atrocities, including deliberate killings of civilians by coalition forces, such as the wanton murder of at least sixteen people by US Army Staff Sgt. Robert Bales in March 2012; airstrikes that slaughtered dozens of Afghan civilians who were celebrating a wedding, traveling in a convoy or simply sleeping; and those killed in small groups or singly at military checkpoints, in firefights and during night raids.

When counting Afghanistan’s dead civilians, it’s useful to break the war down into three phases: the initial campaign, involving a small number of US troops, Special Operations Forces and the CIA, backed by a relentless campaign of airstrikes, in 2001 and 2002; the period from 2003 to 2007, when the Taliban-led insurgency slowly began to gain traction; and the period from 2008 to 2013, which has seen the most intense fighting between the US/ISAF coalition and a mature, resilient insurgency.

2001–2002: During the first months of the war, there was no one to count the dead. The Taliban had fled, and there was essentially no government in Kabul. The United States had almost no forces on the ground, and it wasn’t paying attention to civilian casualties anyway. The UN and NGOs were confined to the capital. Yet many died, mostly as a result of US airstrikes. The Nation’s database, which relies on media reports compiled under extraordinarily challenging conditions, records 136 incidents during the first five months of the conflict, involving between 1,200 and 3,155 war deaths.

Other researchers, adopting a more cautious methodology, came up with somewhat lower estimates. In June 2002, the Los Angeles Times published the results of an intensive investigation into civilian casualties caused by US airstrikes between October 2001 and February 2002. Its reporters visited twenty-five Afghan villages, though most of its analysis was based on media reports. After reviewing more than 2,000 such reports, the paper estimated that between 1,067 and 1,201 civilians perished during that period. Another study compiled by Carl Conetta of the Project on Defense Alternatives, a liberal group based in Washington, also depended on media reports. The study concluded that at least 1,000 civilians—and possibly as many as 1,300 — were killed between October 2001 and January 2002.

But perhaps the most hands-on investigation was conducted by William Arkin, a veteran military analyst and bomb-damage specialist, who visited Afghanistan in 2002 with a team from Human Rights Watch. Though Arkin had often worked closely with the US military, in Afghanistan he got little cooperation from CENTCOM or the Air Force, he told The Nation. Worse, he says, “there was no Afghan partner to work with, no humanitarian organizations, no government that gave a shit or had any records. So there were no records. Even at the local level, there were no records! It was really stunning.”

After identifying hundreds of sites that were targets of US airstrikes and visiting many of them, Arkin says it was impossible to sort out current damage from old wreckage. “It’s not precision warfare on top of a pristine landscape,” he says. “It’s chaos on top of chaos.” He estimates that no more than 1,500 civilians died in the first five months of the war — but, he adds, “absent the military looking at it seriously at the time, which they weren’t; and absent the intelligence community having any responsibility for reporting on this subject, which they didn’t; and absent local government records or tracking, which there wasn’t any of, I defy anyone to say how many people died. We couldn’t.”

2003–2007: We could find no evidence that anyone tried to count the dead during these years. According to The Nation’s compilation, between 617 and 1,012 civilians died in eighty incidents involving the US military and coalition forces during this five-year period, though the actual toll is probably far greater, given the scant attention by all parties.

Immediately after the fall of the Taliban government, the United States and the UN focused almost exclusively on “nation building,” even as the insurgency took root under their noses. This was the era of the so-called “light footprint,” when George W. Bush’s administration was obsessed with Iraq. Only 8,000 US combat troops and a contingent of international forces half that size were in the country, and US ground forces were mostly confined to Kabul and a few big military bases. (It wasn’t until October 2003 that the UN Security Council authorized the expansion of the ISAF mission beyond Kabul.)

“When the insurgency started rearing its head, the way that they fought that was with a lot of airstrikes,” says Sarah Holewinski of the Center for Civilians in Conflict [see Holewinski, page 33, for more on the center]. “They were dropping 2,000-pound bombs instead of 500-pound bombs. The civilian on the ground was not the priority, so quite a lot of civilians were being killed.”

In its 2008 report “‘Troops in Contact’: Airstrikes and Civilian Casualties in Afghanistan,” Human Rights Watch noted: “The combination of light ground forces and overwhelming airpower has become the dominant doctrine of war for the US in Afghanistan. The result has been large numbers of civilian casualties, controversy over the continued use of airpower in Afghanistan, and intense criticism of US and NATO forces by Afghan political leaders and the general public.” Human Rights Watch estimated that civilian casualties from coalition airstrikes rose from 116 people in thirteen bombings in 2006, to 321 people in twenty-two bombings in 2007.

According to Andre-Michel Essoungou, spokesman for the UN’s Department of Peacekeeping Operations, “UNAMA began monitoring civilian casualties systematically in late 2007. It began to systematize the collection and analysis of that data in 2008 when it published its first report on ‘Protection of Civilians in Armed Conflict.’ No organization has good or reliable data on civilian casualties from 2001 to 2006, and we have not seen estimates that would be considered reliable.” The UN’s reports during these years were sporadic and mostly guesswork, and both the UN and the media focused on high-profile, mass-casualty incidents. In 2006, a report by the UN high commissioner for human rights noted that there were “approximately 1,500 civilian deaths in 2005, the highest number of civilian deaths in any year since the fall of the Taliban in 2001.” But the emphasis ought to be on “approximately.” In an early attempt at systematically counting civilian dead for 2007, the UN reported 1,523 killed. In her study, Neta Crawford concludes that, at most, 4,065 civilians died as a result of combat between 2002 and 2007, but she acknowledges that “from 2002 to 2005, there were very few counts or estimates made by independent sources of the number of civilians killed in the conflict.”

2008–2013: By June 2008, 48,250 US troops were in Afghanistan. A year and a half later, after two escalations of the war ordered by President Obama, the US troop level passed 100,000. The insurgency, including the Taliban, the Haqqani group and Gulbuddin Hekmatyar’s Hezb-i Islami, was full blown, and the war was at its peak.

Both the UN and ISAF created formal body-count mechanisms for civilian deaths in 2008, but they were marred by significant flaws, undoubtedly resulting in an undercount. The Nation’s database, which relies primarily on Western media reports and thus also undercounts the dead, reflects 234 incidents, involving between 972 and 2,229 people killed from 2008 to the end of 2012.

According to the annual UNAMA reports, the number of Afghan civilians who died from war-related violence inflicted by all sides rose steadily: from 2,118 in 2008, to 2,412 in 2009, 2,777 in 2010 and 3,021 in 2011, before falling to 2,754 in 2012, with another 1,319 deaths in the first six months of 2013. Of those, the proportion killed by insurgents rose too, from 55 percent in 2008 to 79 percent in 2012. In all, UNAMA concluded, 2,736 of those killed between January 2008 and June 2013 died at the hands of the US/ISAF coalition and Afghan security forces.

Does the US military have better numbers? Probably not. Both ISAF and the US command maintain records of violent incidents as part of a system called the Combined Information Data Network Exchange, which contains more than 100 different kinds of reports tracking battlefield data. CIDNE is classified, and The Nation was not given access even to a sanitized version of it. But researchers who have seen the classified data suggest that the numbers aren’t there, especially before 2008. We asked Larry Lewis, who in 2010 co-wrote a definitive — and still classified — Joint Civilian Casualty Study for the military called “Reducing and Mitigating Civilian Casualties,” if there were any reliable numbers for the war’s early years. “Not that I ever found,” he says, “and believe me, I looked.” One source, who was part of the Civilian Casualty Tracking Cell set up by the military in 2008, says there was a top-secret room containing highly classified data at ISAF command headquarters, adding: “ISAF does keep a log of civilian casualties. But it was classified, of course. In fact, they had the ‘Five Eyes room’ that very few people could go into. It was called the Five Eyes room for the five countries whose senior people were allowed in: the United States, UK, Canada, Australia and New Zealand.”

None of the sources interviewed by The Nation confirmed that either ISAF or the US military has accurate numbers on dead civilians, even after 2008. Lewis says much of the data that the military does have are suspect, because they’ve been unevenly collected. “There are some commanders that, any time a civilian casualty is suspected, they’ll do an investigation to try to get to the bottom of it,” he says. “Others, they’d only do it if they thought there could be neglect or actual criminality. So there are a lot of different criteria.”

Sarah Sewall, who wrote the introduction to the US Army/Marine Corps Counterinsurgency Field Manual, prepared in 2006 under the direction of General Petraeus, was an author with Lewis on the Joint Civilian Casualty Study. Despite the lack of complete and accurate data, she says, the report “relied on a number of anecdotes that we studied in detail, so that’s the way we were able to correct” for the absence of underlying data to draw conclusions about trends and make policy recommendations to the US Army. A nearly complete version of the Joint Civilian Casualty Study was obtained by The Nation.

II. The UN’s Flawed Count

The UN’s human rights and civilian protection team in Afghanistan has done critically important work. Against near-impossible odds, it established a nationwide network of offices and trained personnel to track civilian casualties, investigate incidents, prepare reports, and put pressure on the US/ISAF coalition, the government of Afghanistan and the Taliban to minimize civilian deaths. Since 2001, it has published voluminous reports on the war’s toll, and since 2008 UNAMA has put out annual and semi-annual data on civilian casualties.

However, in a series of wide-ranging interviews, current and former UNAMA officials told The Nation that the UN’s human rights work has been hamstrung by political pressure from top officials to minimize or downplay issues that might undermine the US/ISAF mission; by open clashes with military commanders over whether Afghan deaths should be counted as civilians or combatants; and by severe conditions in which investigations were hampered by security concerns. One former UNAMA official, who spoke to The Nation on background, says that despite enormous problems in collecting data and evidence, the UN’s totals are fairly complete. “I would guesstimate that we’re missing 10 percent of what’s happening out there,” he says. But another former official, who spent years working with UNAMA’s human rights group, says flatly, “What we’ve reported is the tip of the iceberg.”

Especially in the early years of the war, the UN’s effort was very limited. Nazia Hussain of the Open Society Foundations spent much of the 2002–07 period in Afghanistan, including a 2005–07 stint with UNAMA in Kandahar, Jalalabad and Kabul. During that time, she says, security conditions deteriorated sharply, making it harder to go into the field. “There was a lot of confusion and chaos reigning at that time, and civilian casualties were increasing.” She adds, “A lot of the deaths have been attributed to NATO strikes or warlords, but if you double that or triple it, you’ve probably got a realistic number, and it’s probably way more than that.”

UNAMA’s human rights team expanded along with the war. “We have seventy people located throughout the country in the nine regional offices and then the sub-regional offices, which is something like sixteen other offices in addition to the nine,” said one former UN official in 2012. Still, UNAMA was often overwhelmed. The official estimated that in some provinces, there were more than 200 violent incidents per month.

And UN officials are the first to acknowledge that they don’t hear about everything. Often it’s difficult or impossible to visit the site of an incident. Frequently, UNAMA had to interview victims and survivors by telephone, or people from a remote village would have to journey to a provincial capital to visit its office. Often victims of violence weren’t willing or able to share what happened. And sometimes they’d exaggerate the numbers for pecuniary reasons — especially if they thought compensation payments might be offered [for more on such payments, see Turse, “Blood Money: Afghanistan’s Reparations Files”].

A former UN official who spent many years in Afghanistan beginning in the 1990s explains how the organization’s approach changed. “After 9/11 and the Bonn Agreement [of December 2001], the US and the UN started re-establishing itself, and a totally different line was taken on human rights and the impact of war on civilians,” she told The Nation. “Basically, the UN went silent. During the Taliban regime, the UN was all the time talking about things of a human rights nature. And very quickly, in 2001–02, there was a very strong message that the UN was no longer going to do that.” Asked where the pressure was coming from, she says, “I think the UN was seen as being very sensitive to the Washington agenda.” Making matters worse, many traditional allies of the human rights groups and the UN’s civilian protection unit, such as the Canadians, the Norwegians and the Dutch, were part of the US-led military coalition. For members of the coalition, the UN’s job was to build the new Afghan government, not to meddle with human rights issues. “We were told that peace was at hand, and so we had to consolidate the peace,” the former UN official says.

As the fighting grew more intense and the UN ramped up its system for counting civilian deaths, ISAF’s politicization of the count intensified too. Kai Eide, who served as the UN’s special representative for Afghanistan from March 2008 to March 2010, recalled that US officials accused the UN of assisting the insurgency by drawing attention to the coalition’s mistakes. “The UN had a strong human rights mandate,” he told The Nation, “and civilian casualties had increasingly become an issue.” A former UN official who spent years in Afghanistan echoed this, noting, “They’ve become a lot more politicized, so that increasingly there are more and more pressures on UNAMA to check and double-check incidents in which it was alleged that people were killed by pro-government forces, which includes both international forces and Afghan forces. And again, that would have an impact on the tallies.”

Eide recounted an early 2008 meeting with Victoria Nuland, then the US ambassador to NATO, in which he was directly warned about disclosures of civilian casualties. In front of Eide’s staff as well as other US military and civilian officials, Nuland laid down the law: “‘No surprises,” she said sternly,” is how Eide remembers it. “I heard about many of these meetings between our human rights people and those at local offices and lower-level officials at ISAF. I think it was a rather constant effort for quite some time,” he told The Nation. In his 2011 memoir, Power Struggle Over Afghanistan, Eide notes that the “UN could not keep quiet when serious mistakes were committed and caused civilian casualties. Our human rights mandate was clear, and we had no intention of sweeping our concerns under the carpet.”

Further down the chain of command, however, the pressure was intense, and in interviews with The Nation, UN staffers talked about accommodations they’d made to suit US and allied military interests. Often ISAF would claim that any casualties that occurred in a particular incident were combatants, not civilians. Or it would dispute the numbers. Or it would insist that an event UN workers had documented hadn’t happened at all. “So you reach a situation where you have a plausible allegation that something happened in Province X. You go to NATO, and they’d say something like, ‘Well, actually, that’s not what happened,'” says a former UN official, adding that the UN would then reluctantly decide to leave the incident out of its database. Casualty tracking, in effect, became a political negotiation to be resolved by backroom horse-trading. “You’d make a judgment call: ‘OK, well, maybe we’ll throw this one out.'” The former official, who recalls “endless meetings with the military” to decide whether or not to include data, adds, “And with each new report, there was a rival version coming out from NATO saying, ‘Actually, you know, we dispute this, we dispute that, we have other numbers.’”

One of the biggest problems between UNAMA and ISAF in trying to reconcile data was the involvement of the ultra-secret US Special Operations Forces in many civilian deaths. Indeed, according to Lewis and Sewall’s Joint Civilian Casualty Study: “Between 2007 and mid-2009, SOF operations (including SOF-directed airstrikes) caused about half of all US-caused civilian casualties.” Yet for the UN, getting cooperation from SOF commanders was nearly impossible. “With ISAF, it was a bit more open. But with the Special Forces, there was no way we could get any information,” says a former UNAMA official. “They insisted that they had killed insurgents. There was no way in. We’d say they were farmers, because the local population had told us, but there was no entry point.”

The situation was the same within ISAF. Lt. Col. Ewan Cameron of the British Army, who served twenty months in Afghanistan between 2007 and 2009, worked on tracking and reducing civilian casualties. But even with his top-secret clearance, he was totally in the dark about SOF activities in his area of operations. “Where civilian casualties were reported to us that resulted from suspected Special Operations Forces action, we could not corroborate that because we did not know that Special Operations Forces activities happened,” he told The Nation. “Indeed, if we were to ask about it, we would’ve been politely turned off.”

William Arkin agrees: “Once you get the CIA involved, once you get black SOF involved, then it doesn’t matter. They play by different rules.” Arkin says that he’s visited top-secret air operations centers where even the military command itself wasn’t aware of what the Special Operations Forces were doing. “They literally didn’t know,” he says. “I’m telling you, they literally didn’t know.”

III. The Military’s Body Count

When isaf finally set up its mechanism for tracking civilian casualties in 2008, it got off to an inauspicious start. “I had a rude welcome,” says one former ISAF official involved in the early days of the Civilian Casualty Tracking Cell, which was created in 2008 with the purpose of monitoring reports of civilian casualty incidents, logging them into a database, analyzing them and producing reports for the higher command. “Four days after I arrived, the person I was supposed to report to was led out of the office in handcuffs, charged with a breach of the UK Official Secrets Act for leaking information about civilian casualties to Rachel Reid of Human Rights Watch. That incident, unfortunately, set the tone for what would be my year on the job. He was led away in handcuffs for doing essentially what we were supposed to be doing!”

The arrested official, a British Army colonel named Owen McNally, didn’t actually leak any vital secrets. He was, in fact, trying to cooperate with civilian NGOs, and he apparently told Reid about some of the directives and guidelines that the US military and ISAF had issued aimed at minimizing civilian deaths. McNally would eventually be exonerated in Britain, but not before the episode turned sleazy. According to Reid and another woman who did casualty work for a different NGO, the British Defense Ministry tried to tar them both with unfounded rumors that they’d had a sexual relationship with McNally.

“The UK Ministry of Defense fabricated the story that he had given me civilian casualties information, and there was a lot of gutter press around it and wink-wink implications,” Reid says. “What he did give me was some information regarding a change in ISAF’s tactical directives concerning the use of one-source intelligence for raids and military actions. The military was very angry, because it was implicit recognition that they’d been conducting airstrikes based on one source.” The scurrilous stories appeared in several British papers, including The Sun, The Times and the Daily Mail.

That troubled start to the Civilian Casualty Tracking Cell was never completely overcome, although in time ISAF and the US command would earn grudging respect from the UN and some NGOs—including Human Rights Watch and the Center for Civilians in Conflict—on issues related to civilian casualties. Still, the Civilian Casualty Tracking Cell and its successor units didn’t always play nice with the UN, and they resisted cooperation with the media, including The Nation.

As noted, in the early years of the war civilian casualty issues were apparently all but ignored by the US military and its coalition partners. A search by ISAF personnel of their records revealed not a single study or survey of civilian casualties between 2001 and 2008. As John Bohannon, who researched the issue for the journal Science, noted in 2011, “The organization in the best position to directly record civilian casualties is the military itself, with nearly 150,000 observers on the ground witnessing the violence every day. But it seemed that the military kept no record of those observations.”

It wasn’t until 2005 that, according to the Army’s Afghanistan Civilian Casualty Prevention handbook, civilian casualties “became a key operational issue in Afghanistan.” And these efforts, it turned out, were failures. “Despite efforts to reduce civilian harm caused by coalition forces,” the 2012 manual reads, “initial initiatives in Afghanistan were not successful in mitigating the issue.” Only after a series of attacks that led to mass casualties did the US military finally start restricting the use of force.

In 2007, Gen. Dan McNeill issued the inaugural ISAF tactical directive. Focused on more effectively shielding civilians during “raids, pre-assault or preparatory fires, and air-to-ground or indirect fires,” the document called for using small arms instead of airstrikes whenever feasible, and limited attacks on compounds to situations in which coalition forces “were taking fire from the compound or there was an imminent threat from the compound, and when there were no other options available to the ground force commander to protect the force and accomplish the mission.”

Still, according to Col. John Agoglia, who headed the Counterinsurgency Training Center in Kabul from 2008 to 2010, the message didn’t resonate. Before 2008, he told The Nation, “there was an awareness, but that awareness wasn’t getting across—in the base, in the field, at the training center.” Similarly, Larry Lewis, who carried out a study of tactical directives for the Pentagon, told The Nation that while ISAF did recognize problems with its methods, the ad hoc response had a limited effect. “So you can see the command was already acknowledging this is detrimental to the mission back in 2007,” he says. “But they were trying to do it in just kind of a ‘Hey, we know this is bad, so we better put out some guidance and hopefully that will help.’ And it really didn’t help.”

In September 2008, in the wake of a US bombing that killed as many as ninety-two Afghan civilians in the village of Azizabad in Herat Province [see Dreyfuss, “Mass-Casualty Attacks in the Afghan War,” at TheNation.com for more], Gen. David McKiernan issued a new ISAF tactical directive that superseded McNeill’s 2007 document. The new order put in place more stringent guidelines for airstrikes, emphasizing the need to reduce civilian casualties during “escalation of force” procedures. Additionally, it “called for acknowledgement of civilian casualties or property damage at all levels, from community level to national level,” and created a mechanism to document civilian casualty incidents through bomb-damage assessments. Three months later, McKiernan issued a slightly revised version that included more restrictive language on airstrikes. The Azizabad attack also prompted CENTCOM to issue a tactical directive to speed the investigation and reporting of civilian casualty incidents.

In May 2009, another devastating US airstrike, this time in Farah Province, killed as many as 140 civilians, according to an Afghan government inquiry. In its wake, CENTCOM commissioned a study to “analyze incidents that led to coalition-caused civilian casualties,” according to a formerly secret briefing obtained by The Nation via the Freedom of Information Act. It found that the 2008 ISAF directive “could have mitigated impact of the Farah incident” but was evidently not followed. The findings also called into question just “how institutionalized this tactical directive” was and found that the rules of engagement were not up-to-date and synchronized.

The study focused a great deal of attention on “fighting the information war.” Briefing materials note that while the International Committee of the Red Cross and Al Jazeera were on the scene of the Farah attack in twenty-four hours, it took three days for the first coalition representatives to arrive and sixteen days for ISAF to issue a press release. “As a result,” reads the secret briefing, “first impressions in the media were established by the Government of Afghanistan, local community interests, and possibly by the Taliban.” ISAF, the study concluded, had ceded “the narrative to those whose interests did not align” with its own. According to a closing summary, the secret report called on ISAF to “engage in the battle for the narrative” in order to minimize the public relations fallout.

General McChrystal issued another tactical directive in July 2009, and in 2010 ISAF issued directives to discontinue the practice of firing warning shots, to limit night raids and to drive in a more courteous manner. “Whereas before, the rules were focused on the problem we had, which was dropping bombs on residential compounds, now they’re focused on any area where there might be a civilian,” said then-Col. Rich Gross, McChrystal’s chief legal adviser (now a general himself). At the same time, the United States was about to lose another battle in controlling the narrative: Pfc. Bradley (now Chelsea) Manning was about to put the lie to the military’s long-held refrain that “we don’t do body counts.”

In July 2010, military documents made public by WikiLeaks demonstrated that the US military secretly maintained files relating to 4,024 Afghan civilian war deaths between January 2004 and December 2009 (also known as the Afghan War Logs). For the first time, the public was privy to secret, internal military reports—unprocessed intelligence from troops in the field—detailing the carnage occurring across Afghanistan.

While suggestive, the documents included only cases from the military’s raw operational reports. Larry Lewis said his investigation of civilian casualty incidents from 2001 to 2008 found very incomplete data. “There’s really nothing good out there,” he told The Nation. “I have tried to pull together some numbers that go back to 2007, but I will say…they’re wrong. They’re definitely incomplete, because all I could do was to go to US Army investigations that were initiated because of civilian casualties and use that as a source. But I’m certain there were other incidents that never had an investigation.”

Separately, the Civilian Casualty Tracking Cell was by now logging data on noncombatants killed and wounded by coalition forces from reports sent in by units in the field to ISAF Joint Command. But these data, too, were “inconsistent in type and quality,” according to Lewis and Sewall’s Joint Civilian Casualty Study. As late as February 2010, in fact, the commander of ISAF “was still asking straightforward questions such as whether US-caused CIVCAS [civilian casualty] incidents could be correlated with particular units and/or with that unit’s length of time in theater, and not receiving answers.”

In 2009, ISAF created another investigative body, the Joint Incident Assessment Team, which deploys to sites of reported civilian casualty incidents to conduct separate inquiries parallel to ISAF’s standard investigations. “A JIAT is used to quickly determine the facts when we have an allegation of a significant event such as a civilian casualty,” Air Commodore Michael Wigston of the British Royal Air Force, former director of air operations for ISAF Joint Command, explained to The Nation. The Joint Incident Assessment Team, composed primarily of ISAF personnel with two “Afghan partners,” then compiles a “factual narrative of events…based on interviews with people involved in the alleged incident.” It is, however, only a fact-finding group and is specifically directed to avoid “any issues that are the purview of a formal investigation.” Like other ISAF investigation teams, JIAT reports incorporate no independent experts, receive no outside oversight and are not publicly released.

In 2011, ISAF created the Civilian Casualty Mitigation Team, with a mandate to provide ISAF leaders with “strategic assessments and recommendations to prevent and mitigate CIVCAS.” With a mission statement asserting that the command “takes every effort to prevent and if necessary assess and mitigate each and every CIVCAS event,” ISAF built what it calls a “CIVCAS community,” including the Civilian Casualty Tracking Cell, the Civilian Casualty Mitigation Team, the CIVCAS Mitigation Working Group (which holds monthly meetings with Afghan leaders to discuss relevant issues) and the NGO CIVCAS Working Group (which does the same with NGOs). Methods of investigation were also refined and systematized.

As it developed and expanded, the CIVCAS community largely operated behind closed doors until—in the wake of the WikiLeaks disclosures—ISAF uncharacteristically opened its civilian casualty tracking operation to Science’s John Bohannon. “They were very forthcoming,” Bohannon told The Nation. “They would always say, ‘Look at this, look at that,’ always with an eye toward reducing casualties.”

After some negotiation, Bohannon was granted access to the data, publishing an article in Science in 2011 and making portions of ISAF’s figures available online. “Our database is 100 percent transparent,” said US Navy Rear Adm. Gregory Smith, NATO’s director of communications in Kabul at the time. But ISAF released data on civilian casualties only by region and month, not on specific events. “I pushed them absolutely as far as I possibly could,” Bohannon says. “What they offered up front, at the beginning, was even less detailed and extensive.”

Just before publication, jittery high-ranking military officials asked Bohannon to postpone the story, he told The Nation, but his article was published without delay and, he later heard, was well received in military circles. Within a year, however, ISAF would cut off his access to the CIVCAS community and its data, while providing less than satisfying answers about the change in policy. “Now you’re back in this Kafka castle,” Bohannon says. “You don’t even get reasonable answers. You don’t get coherent answers.”

When The Nation requested an embed with ISAF forces in order to witness the military’s tracking system in action, we were repeatedly rebuffed. Despite the fact that Bohannon had been given access months earlier, The Nation was told that since the Civilian Casualty Mitigation Team and Civilian Casualty Tracking Cell work inside a “secure facility,” members of the public were barred. The Joint Incident Assessment Team commander gave a similar response to an embed request. An ISAF spokesman told The Nation that he “declined to have anyone embed with him on investigations since some of the information is classified.”

Official data indicate that ISAF’s much-vaunted efforts to avoid civilian casualties have had, at best, limited results. Internal reports mistakenly released to The Nation by ISAF indicate that for the first three- quarters of 2011, ISAF forces were responsible for 434 civilians killed or wounded, up from 414 during the first three-quarters of 2010, while deaths attributed to coalition forces ticked only slightly lower, from 175 to 166. Whether this is an artifact of better surveillance and tracking procedures or a failure of ISAF policies is unknown, and it remains unknowable because of ISAF’s veil of secrecy.

Various investigations, including Bohannon’s report in Science and Neta Crawford’s “Costs of War” project, have noted that the US military’s database misses even mass-casualty incidents, such as the September 2009 massacre in the Ali Abad district of Kunduz Province that left nearly 100 people dead, mostly civilians who’d gathered around stalled tanker trucks to collect fuel. In her report, Crawford says: “NATO forces eventually acknowledged that most of those killed were civilians, and Germany made condolence payments to the families of 91 civilians killed and to the families of 11 wounded. Yet, the ISAF CIVCAS database does not record any civilian deaths due to close air support for September 2009 in northern Afghanistan.”

How could the military miss scores of dead and not include them in its own database, even in a widely publicized case such as Kunduz? Clearly, someone in the military bureaucracy is unwilling to admit that the people slaughtered were civilians. Which raises the question: How useful is the Civilian Casualty Tracking Cell database?

IV. Lessons Learned?

The best way to prevent civilian casualties in war is, of course, to avoid war. Short of that, perhaps the best that can be hoped for is that the Defense Department and the military command learn the right lessons from the war in Afghanistan. However, there’s little evidence that these lessons are being institutionalized. And some may have learned the wrong lessons, such as the illusion that the widespread use of “precision” drone missile attacks can reduce civilian casualties. This ignores the untold number of innocents killed in such strikes (the Bureau of Investigative Journalism estimates up to 1,000 in Pakistan, Yemen and Somalia alone), which has created a new generation of anti-American fighters seeking revenge for loved ones and comrades killed.

Sarah Holewinski, executive director of the Center for Civilians in Conflict, has spent years training US military officers on how to avoid civilian deaths, and she’s worked with US and ISAF commanders in shaping directives for the troops. While she says that many of those with experience in Afghanistan did learn the right lessons, it’s far from clear that the Pentagon is preparing to apply them broadly. “What’s needed is an Office of Civilian Protection,” Holewinski says. “You really should have at least a focal person, if not a team, saying, ‘What have we learned? Where in our policies and protocols, and new procedures, and new counterterrorism strategy, can we put these understandings on preventing civilian harm?’ And we’ve been pushing that for about five years, and it really hasn’t gone anywhere. A lot of people I’ve talked to say it’s a bridge too far; they don’t have the resources. Everyone knows civilian casualties are important, but that’s not enough. It doesn’t mean things will actually change.”

The Joint Civilian Casualty Study acknowledges that, as of 2010, there was no Pentagon office that directly focused on civilian casualties. Further, “there is no cadre of ‘experts’ on civilian casualties and U.S. military operations, nor is there an existing body of knowledge on the topic.” Partly this is because of sheer neglect, and partly it’s because personnel—including senior officers—retire or change jobs and take what they’ve learned with them.

When The Nation asked the study’s authors if there was now, finally, an office in the Defense Department concerned with civilian casualties, the response was a suggestion that it existed in the Office of the Secretary of Defense. But nothing turned up. One officer said, ”I called around to all the offices I could think of in OSD, and they all responded, ‘Not me! Not me! Not here.’” Inquiries elsewhere got similar results.

As the war in Afghanistan winds down, the American people, the media, academia and think tanks all have a role to play in demanding that in any future wars, the United States place the highest priority on avoiding civilian casualties and, if they occur, on being accountable and making amends. If the Pentagon moves slowly, the quickest route is for Congress to hold hearings and then write legislation creating and generously funding such an office, and insisting that its procedures be codified. That, at the very least, would begin to give meaning to the deaths of tens of thousands of Afghans who perished in a needless, misguided and horribly run war.

Click here to view the interactive database of America’s Afghan Victims.

This article was reported in partnership with The Investigative Fund at The Nation Institute, now known as Type Investigations.