The US military strike that devastated the MSF hospital in Kunduz in northern Afghanistan on 3 October generated profound, if short-lived, outrage in much of the world. The US government promised an investigation, and in late October appointed a military panel to do so. Yet its investigation is unlikely to address the more fundamental questions this attack raises: Why have US-airstrikes repeatedly produced catastrophic cases of “collateral damage” in Afghanistan? How, if at all, can the civilian impact of such means and methods of warfare be reduced – not only in the continued conflict Afghanistan, but also in similar airstrikes in Iraq and Syria?

Armed conflict in Afghanistan, Iraq and Syria during the past decade has inflicted tremendous cost on civilians. Some of the deaths and injuries have been the direct, but unintended, consequences of specific Western military operations, mainly the United States.  While accounting for a relatively small part of the total number of civilian casualties, such incidents are significant in a broader logic that seeks to understand if and how these kinds of costs of war can be minimized.  Two aspects of US military operations are particular relevant here. As the most advanced military power of our age, the US has a highly developed technological capacity to use “smart” weapons that discriminate between military and civilian targets. Moreover, the US is a signatory to the principal corpus of international humanitarian law, and successive governments have accepted customary principles of such law. Nevertheless, incidents of severe “collateral damage” occur. Why?

A project supported by the Research Council of Norway to explore broader issues of protection of civilians in contemporary conflict has opened a door to examine the impact on civilians when US and allied airpower is applied in a conflict, often for the ostensible purpose of improving the lives of the people. A case study of recent experiences from Afghanistan found variable reasons for why so-called mistakes occurred, but systemic patterns of subsequent response from the military that suggested little improvement for the future.  Three major cases and the general dynamic that emerges are discussed below.

The main conclusions are:  

  • The US military in Afghanistan did change procedures to reduce unintended civilian causalties from airstrikes. The principal reason was a change in military strategy that prioritized “winning hearts and minds” of the local population;
  • External vigilance through monitoring and independent investigation had an impact, but in itself did not lead to the change in “system design” in operations that altered the balance of risk in favour of civilians.

The principal policy implications follow:

  • A significant risk of civilian casualties is inherent in the use of airpower in asymmetric conflicts.
  • While external monitoring remains important, mobilization for alternatives to military intervention of this kind is necessary to address the underlying causes.

* * * * *

Two questions

In the recent bombing in Kunduz, the MSF immediately condemned the strike as a grave violation of international humanitarian law, and possibly a war crime. The organization also demanded a full, transparent and independent investigation, but, so far, the US military has decided it will investigate itself.

Previous cases of large civilian casualties from US airstrikes in Afghanistan have been investigated by the US military as well as several external bodies – the UN, the Afghan Independent Human Rights Commission, the International Committee of the Red Cross, Afghan government commissions, and sometimes joint commissions of the Afghan government and the international military command (ISAF). None has been transparent, independent and full.

External investigations have mostly been transparent and independent, but lacked critical data to fully assess whether, in a particular case, the rules of war had been followed. Did the military units involved take all feasible precaution to discriminate civilians from combatants? Did they only use force that was proportionate and necessary in relation to the military objective to be achieved? The US military had data bearing on these questions, but mostly kept it for internal use. Their own investigations, in turn, may or may not have been full; we don’t know because they were not transparent. Obviously they were not independent, and snippets of reports that have been declassified show a pattern of denials and great elasticity in interpreting the rules of war.

Three cases

Consider 3 well-known cases. For a fuller analysis click here.

The Uruzgan wedding party

Two large, extended families were celebrating a wedding in the cool, early night of 1 July, 2002 in the southern province of Uruzgan. As is local custom, the men fired guns in the air to mark the occasion. The response from the air was ferocious. A plane over the village started shooting and pursued villagers who scattered to seek shelter. After the plane had left, American and Afghan soldiers arrived. They bound the hands of survivors, took pictures of the dead and the living, and removed shrapnel and bullets from the corpses. Next on the scene was a UN team that arrived within 48 hours from nearby Kandahar. Then came journalists to interview survivors and local officials.

There were, in short, many witnesses, and several investigations followed. At the request of the Americans, the report of the UN team was not made public, but was said to have found 89 persons had been killed and up to 200 injured – all civilians. An Afghan government report had a lower number, but likewise found that all the victims were civilian. The high casualty figure was not surprising considering that the plane in question was an AC-130, equipped for “saturation-fire”.

The American military investigators confirmed that all known victims were civilians - 34 dead and 50 wounded – but maintained the attack had not been a mistake. Taliban had been operating in the village and fired at the US plane, their report claimed, and the gunship fired in response. The dead and injured were simply collateral damage. The report admitted that the pilot did not actually see the anti-aircraft weapons on the ground –but noted that the area was a Taliban stronghold and enemy forces had fired at planes from the same site two days previously. Hence, any fire coming from the area was presumed to be hostile, and the responsibility for the dead and injured rested with the Taliban who operated in the midst of a village.  None of the villagers interviewed by independent journalists and Afghan officials supported the claim of anti-aircraft fire that night, but it was their word against the US military, and the UN was muzzled. More disturbingly, the military report noted that just as the pilot could not actually see the anti-aircraft sites on the ground,” it is also not possible to distinguish men from women or adults from children.” The clear implication was that discriminating civilians from combatants – a fundamental principle of international law - likewise was not possible. Leaving that thought aside, however, the military investigators concluded that the gunship pilot acted “properly and in accordance with the rules”.

The attack on Azizabad village

Six years and several hundred civilian casualties produced by US airstrikes later, another devastating and controversial attack took place in the western province of Herat. US forces had been informed that the Taliban were preparing a large meeting in the small village of Azizabad, and attacked with their own and Afghan forces, assisted by close air support. When the dust cleared, around 90 civilians – the majority children - were dead, and many more injured, according to investigations by the UN, Afghan NGOs, international NGOs and the Red Cross. Most had been killed by airstrikes.

The villagers maintained there were no Taliban in the village. Local rivals had planted false rumours.  Independent investigations gave credence to this information. The report issued by the US Central Command (CENTCOM) dismissed the independent investigations as mere allegations based on hearsay. For their own part, the US military maintained they had solid intelligence that the Taliban were assembling in the village. Even if this had been the case, was bombing the village and thereby failing to discriminate civilians from combatants an appropriate response? The independent investigations said no, the US military investigation said yes, claiming the response was “proportionate’, “necessary” and in compliance with the laws of war. The military count of civilians, moreover, was much lower (33 civilians and 22 insurgents killed).

The bombing of Garani village

A few months later US planes bombed another village in neighboring Farah province. This time they dropped five 500-pound and two 2000-pound bombs – enough to destroy a couple of airfields. The UN and the Afghan human rights commission counted 80 civilian deaths. A subsequent investigation by CENTCOM determined that only 26 civilians were killed, while the remaining 78 were insurgents. This time there was agreement there had been fighting in the area; in fact, US and Afghan forces had called in air support while waiting to evacuate two of their wounded. The bombers had arrived at nightfall and the crew had spotted suspicious movements in the area, “similarly-sized adults moving in a tactical manner – definitively and rapidly in evenly spaced intervals across difficult terrain the dark”, the CENTCOM report wrote. Concluding that they were Taliban fighters heading for shelter in a building, the planes attacked. The version told by villagers differed: The Taliban had left the village by the time the planes arrived, and the people had been running to the mosque for shelter. 

Unlike earlier reports, this CENTCOM investigation opened for some criticism and made a small admission; the “inability to discern the presence of civilians and assess the potential collateral damage of those strikes is inconsistent with the U.S. Government’s objective of providing security and safety for the Afghan people.” The ratio of military lives saved (2) to civilian lives lost (80) also raised questions about proportionality. Yet, the report concluded, “the totality of the circumstances…validated the lawful military nature of the strikes”.  

What went wrong?

As we have seen, things went wrong on several levels, starting with the context. In the first case, the atmosphere of revenge in the US after 9/11 had permeated the military operational culture as well. As a US pilot later said, it was a “fangs-out, kill-kill-kill culture… the mind set was: maximum number of enemy killed, maximum number of bombs on deck, to achieve a maximum psychological effect.” In a well-known “Taliban-country” like Uruzgan, almost anyone was a likely target.

In the second incident, the familiar mechanism of conflict manipulation was at work. Local conflicts over land, water, and a myriad of feuds over the past decades of Afghanistan’s violent history had shown a tendency to attach themselves to the overarching political cleavages of the day. In the early 2000s, the principal divide was Taliban vs the government, assisted by the US-led coalition. Branding a rival as a Taliban was a sure way to focus the attention of the international military. More problematic was the readiness of a faction to put on a Taliban hat – at least for the time being, just as a rival faction would put on a government hat, when that paid divided. The hats were typically changeable and in themselves had little meaning, but they were useful assets in local conflicts. Western military intelligence might be aware of this dynamic,*1 and ask if there were other ways of dealing with local conflicts that had little strategic relevance outside the local Afghan arena. In the operational codex of the US-led coalition, however, Taliban-hatted persons were obvious targets.

In the third case, the value assigned to the lives of “our” soldiers overrode instructions to take particular care to identify targets, issued by the US command of the international military forces in Afghanistan after earlier cases of catastrophic “collateral damage”.

There was also a systemic problem. Dropping high-tonnage bombs or using planes equipped for “saturation-fire” in densely populated areas makes it virtually impossible to discriminate between civilians and combatants, as required by law. As Neta Crawford has argued, civilian casualties in such circumstances are neither accidental, nor a “tragic incident”, as often described by the military.*2 They are “normal accidents”, to borrow from the concept developed by Charles Perrow.*3

Using “smart” weapons that make it possible to discriminate the target more carefully is a way of altering operational design that would reduce civilian casualties. The emphasis on drones and other weapons that can discriminate more readily illustrates such changes. Yet, as the MSF case demonstrates, use of airstrikes in populated areas against an enemy that fights from, in, and often is part of, the local population has an inherent probability of causing large civilian casualties. Evidence to this effect has been carefully tallied and is accumulating.

The issue goes to the heart of the nature of contemporary conflicts when military forces equipped with extremely advanced technology are fighting against militias or irregular forces that operate in populate areas and easily melt in among civilians. In asymmetric conflicts of this kind, states that possess advanced technology will obviously use that even if it causes civilian casualties that, in principle, they are concerned to keep down. The logic is particularly compelling for states intervening in what at home can appear as “foreign wars of choice”, and whose governments are under domestic political pressure to minimize their own casualties. Hence the reliance by the US and its allies on airpower in the recent phase of the war in Afghanistan, Iraq and Syria.

What can be done?

In the case of Afghanistan, something was done to increase protection of civilians against US-led airstrikes. As civilian casualties mounted, the UN mission in the country (UNAMA) engaged itself, alongside the Afghan government and NGOs to demand that the international military forces in the country (ISAF) institute better procedures to monitor and investigate the effects of their operations on civilians. The ISAF command recognized the need to give a little, and introduced some reforms along this line. But the principal change came with adoption a new military strategy that emphasized the need to “win the hearts and the minds” of the Afghan people. The new counter-insurgency doctrine emphasized by General Stanley McChrystal in 2009 came with detailed instructions to the forces to limit civilian casualties even if it meant greater risk to themselves.

Civilian casualties produced by US airstrikes shrank immediately, despite an increase in operations. Further evidence that the orders were taken aboard at the field level came from serving US soldiers who complained loudly to the press and to their representatives in the US Congress that the guidelines, which in particular tightened the restrictions on calling in close air support, increased the risk to their own lives and impaired their fighting effectiveness. Some US military lawyers argued that the new directives unduly shifted the balance of risk in combat areas to favour civilians.


The counter-insurgency doctrine did not last long in Afghanistan, has been all but abandoned by US and its allies in Iraq, and was never applied in Syria. Given the recognized difficulties of fighting a counter-insurgency in a faraway country – as demonstrated most recently in Afghanistan – Western military establishments seem unlikely champions of strategies to protect civilians in these kinds of conflicts.

That shifts the focus to other approaches to limit the means and methods of warfare and improve monitoring and accountability. The government of Norway organized in 2009 a 4-year rolling set of conferences to “reclaim the protection of civilians under international humanitarian law”. Concerned NGOs have argued for formal restrictions against the use of bombs in populated areas. The ICRC has proposed stricter guidelines for who is a non-combatant. In the Kunduz case, MSF wants to reactivate an international humanitarian fact-finding commission that was established in international law in 1929, was confirmed in additional international law in 1977,  actually took form as a body of commissioners with a budget and a secretariat in 1991, but has never functioned. The list continues.

The problem is not primarily the lack of legal instruments that define norms and institutions of accountability. The UN has on several occasions established ad hoc fact-finding commission to investigate possible violations of international humanitarian law.*4 But even less powerful states than the United States are notoriously difficult to hold to account. The other problem lies in the nature of the violation. Civilian casualties that are not intended, but an incidental consequence of military operations – often called “a mistake”- fall in a different legal category from civilian deaths that are willfully and intentionally caused. To establish that a “mistake” violates international law, it must be demonstrated that the military action was not proportionate or necessary in relation to the military objective to be achieved. This requires, as noted, a great deal of data that the military command alone controls. The very concepts of “proportionality” and “necessity”, moreover, are open to interpretation.*5 Very few cases of “collateral damage” from Afghanistan have even reached a judicial inquiry, of these, reduction in rank or dismissal from the force has been the usual outcome. The case of the German colonel who in 2009 called in a US airstrike that killed around 140 persons near a village in Kunduz province is a case in point. German prosecutors dropped the proceedings against Klein on the ground that he was not in a position to determine that the persons in the area were civilians and not insurgents.

Implications for action

The point, as Neta Crawford forcefully agues in her book, is that even if unintended civilian casualties from modern warfare can be very high, they are not easily limited by recourse through law. The added point, stressed here, is that given the nature of contemporary conflict, military actions by the US and its allies are likely to produce precisely this form of civilian casualties when they intervene in third countries – whether with drones or manned aircraft. In Afghanistan, it took external monitoring and pressure combined with powerful strategic incentives within the military to significantly reduce “collateral damage”. In other countries where the US and its allies currently are involved – Syria and Iraq - these conditions do not obtain.  There is very little access for external and independent monitoring of Western airstrikes, including procedures for targeting, and the effects on civilians. The US drone campaign in northwestern Pakistan is slightly more open to external monitoring, which has found civilian casualties to be markedly higher than those claimed by the US military and CIA.  And whatever the military strategic rationale for US intervention in Syria and Iraq is, “winning hearts and minds” in a counter-insurgency campaign is not among them.

Where does that leave individuals or governments concerned to protect civilians against US-led airstrikes? There seems to be three alternatives, in ascending order of radicalism. The path of least resistance would be to accept collateral damage as a systemic and inherent dimension of external military interventions of this kind.  “Stuff happens”, as then Secretary of Defense Donald Rumsfeld said when Baghdad was looted in the wake of the American invasion in 2003. A more active approach would stress political action to improve monitoring and accountability and change “system design” through better targeting and related measures. Yet the prospect for change is subject to structural limitations. Military establishments are disinclined to allow external monitoring of their operations by sharing information, as the UN mission in Afghanistan experienced. Altering the balance of risk in the system design to favour civilians comes with a cost – the increase in risk to own forces or expensive aircraft. The third alternative is to employ means of statecraft other than military force to manage conflicts in weak or collapsing states. Political solutions, of course, are not easily found, but the cost of war as demonstrated by the difficulty of avoiding “normal accidents” tips the scale at least a bit further in favour of the saliency of diplomacy over military force.



*1 Mike Martin, An Intimate War. An oral history of the Helmand Conflict. London: Hurst&Co, 2014

*2 Neta Crawford, Accountability for Killing. New York: Oxford University Press, 2013.

*3 The concept of «normal accident» as developed by Charles Perrow designates an operation where negative consequences are unintended, yet probable and foreseeable because they flow from the operational design. Charles Perrow, Normal Accidents: Living with High Risk Technologies. Princeton, NJ: Princeton University Press, 1999.

*4 Well-known examples are the fact-finding commissions established by the UN Human Rights Council to investigate possible violations of international humanitarian law duringr Israel’s war against Palestine in 2009 and 2014.

*5 See e.g. Michael Newton and Larry May, Proportionality in International Law, New York: Oxford University Press, 2014.