Attacks and Warnings in Urban Warfare

by | Nov 9, 2023

Urban warfare

The media coverage of recent armed conflicts has brought to a wider audience the existential nature of urban warfare and its humanitarian costs. For some, the era of urban warfare has arrived. From an historical perspective, however, there is nothing new about fighting in populated areas. Cities have always been transmitters of war (Mumford, The City in History, p. 46). For an attacking force, this is because cities and their civilian populations are the primary source of an adversary’s political and economic power. Degrade that power and you undermine the resilience of your opponent to make war (Di Marco, Concrete Hell, p. 16).

At the same time, fighting within a city is often advantageous for its defenders. Densely populated areas allow fighters to hide in plain sight amongst the complex physical terrain while securing support and resources from a sympathetic civilian population. This is especially important for organised armed groups that seek to negate the intelligence collection capabilities of conventional armed forces.

Simply put, experience teaches that military control over cities is a recurring and unavoidable imperative during armed conflict. As such, the victims of modern urban warfare have much in common with those who endured the historic hardships of siege warfare.

These hardships help explain the development of the law of war rule concerning the advance warning of attacks against populated areas. Early examples of State practice in this regard include the Lieber Code (art. 19) and the Brussels Declaration (1874), which outlined the expectation that commanders, whenever admissible, would do all in their power to warn the authorities before attacking a defended town. The 1907 Hague Regulations for Land Warfare later codified this customary practice as treaty law. Today, Additional Protocol I (AP I), Article 57 provides the broader regulatory framework for the taking of precautions in attack as a means of affording general protection to the civilian population. Within this framework, Article 57(2)(c) requires that “effective advance warning shall be given of attacks which may affect the civilian population, unless circumstances do not permit.” States generally accept this rule as reflecting customary international law.

Ahead of the Second Battle of Fallujah (Iraq, 2004) and the Battle of Marawi (Philippines, 2018), the attacking forces gave such a warning of their impending assault of the respective cities. This presented two key advantages. First, it allowed a significant portion of the civilian population to evacuate the city, thus reducing the incidence of civilian death and injury caused by the subsequent military operations. Second, the reduced presence of civilians provided the attacking force with greater freedom of manoeuvre and simplified the task of distinguishing military objectives. Thus, advance warning served both a humanitarian and military purpose. By contrast it was more difficult to give advance warning of individual attacks once combatants were committed to the city.

Against this backdrop, this post examines two specific aspects of the requirement to give advance warning of attacks. It begins by analysing the definition of attacks. It then considers the circumstances that might not permit a warning. The post concludes it can be difficult to assess where an attack begins and concludes in the urban environment. Further, the same circumstances often preclude troops engaged in ground combat from giving effective advance warning. In view of these practicalities, giving advance notice to the population ahead of a large-scale military operation is often the only effective way of providing civilians a chance to protect themselves. Notice also offers the adversary an opportunity to discharge its obligation to take precautions against the effects of an attack, for instance, by removing its forces from densely populated areas.


The obligation to give effective advance warning applies only to attacks. In this context, an attack is a narrow concept and a distinct subset of military operations. AP I, Article 49 defines attacks as “acts of violence against the adversary, whether in offence or in defence.”

Three aspects of this definition require elaboration. First, an attack is a technical term, which is limited in time and space. An attack is thus a conscious course of action—a specific military operation—decided upon by combatants because of the operational situation in which they find themselves. Second, an attack is distinct from the jus ad bellum concept of armed attack (UN Charter, art. 51). The instigator of an attack is the party that launches combat action to target an adversary within the broader construct of an armed conflict, irrespective of which party was the initial aggressor (International Committee of the Red Cross (ICRC) Commentary, para. 1880). Finally, the inherent characteristic of an attack is the projection of violence. The primary objective of war is the submission of enemy forces. The inescapable motive of combat action is the infliction of violence upon others and their property. It follows that to succeed in war it is necessary to coerce the enemy, wherever the enemy exists. An attacking force can only achieve this by projecting force into areas beyond its immediate and effective control.

In many battlefield situations, the identification of an attack is straightforward. The use of artillery to bombard an enemy trench position will constitute an attack. An air strike that uses explosive force to target an enemy military command headquarters located within a city likewise involves the projection of violence and is consequently an attack. However, the frenetic nature of urban warfighting by ground troops makes it more difficult to assess when a specific military operation involving the use of arms begins and ends, and, accordingly, when the legal obligation to give effective advance warning arises.

Consider the use of explosives by a dismounted infantry patrol within a built-up area to clear obstacles that prevent their movement through the cluttered streets or to gain access to buildings and compounds. On one reading it is possible to argue that the use of explosives to blow a hole in a wall through which the soldiers can then move comprises an attack. This is because the act of firing the explosives involves the deliberate use of force for the specific purpose of damaging or destroying property, namely, the wall. Yet in this example, the wall and the area immediately surrounding it lie within the effective control of the troops. As such the use of explosives to breach the wall does not involve the projection of violence against the enemy.

The issue for the on-scene commander is not, therefore, whether the use of explosives involves an attack but whether breaching the wall is a necessary interaction with their immediate environment in much the same way as a military force on a rural battlefield might have to penetrate dense hedgerows to maintain momentum. An important distinction is that even though the use of explosives to remove the obstacle is not an attack, it is undeniably a military operation. The troops involved must accordingly take constant care to spare civilians and civilian objects. Nonetheless, the obligation to give effective advance warning does not arise.

Another challenge at the tactical level concerns determining whether a specific military operation constitutes a single attack or multiple discrete attacks in respect of which the armed force must give effective advance warning. If the operational commander orders the destruction of an urban military command headquarters, the resulting military operation will be an attack irrespective of whether an aircraft or infantry force is allocated to the task. Just as poor weather conditions may preclude the aircraft from completing its mission, however, so might the fog of urban complexities enshroud the advancing ground troops.

For instance, as the soldiers approach the environs of the command headquarters, they could receive fresh orders to destroy an armoured vehicle that has unexpectedly moved into the area to defend the original target. The question arises whether destroying that armoured vehicle involves a discrete precursory attack that requires the reappraisal of precautionary measures or whether the new target forms an inextricable part of the original pre-planned attack. Likewise, if the infiltrating force is the subject of an ambush as it approaches the target, any return of defensive fire will not amount to a separate attack that requires warning because the adversary is the immediate initiator of that combat engagement.

An even greater challenge will arise if the infiltrating force encounters an enemy that decides to “shoot and scoot” amidst the city, firing upon the soldiers before temporarily withdrawing and then relocating to fire upon them again. Such circumstances make it far more difficult to ascertain the stopping point of one attack and the start of another, a reality that the deployment of other infantry teams pursuing different lines of military effort will compound.

These practicalities highlight that the bounded concept of attack, the existence of which is a precondition of the obligation to warn, might be simple to apply in the context of an overarching operation to assault a city or deliberate pre-planned targeting operations conducted by aircraft. Yet the parameters of an attack are more difficult to discern when the fluidity and tempo of tactical activity cause a series of individual combat actions to bleed into one. Such activity is the mainstay of urban warfare. It may not even occur to an on-scene commander in those circumstances that their troops are embarking on the formalities of an attack for which there is an obligation to consider the giving of advance warning.

When Circumstances Do Not Permit

Assuming the urban complexities allow a commander to ascertain the starting point of an attack, the obligation to give effective warning is not absolute. The attacking force need not warn if circumstances do not permit. The use of this qualifier stands in contrast to other provisions of AP I, Article 57 that use the term feasible, with the result that commanders have a wider discretion when determining the practicalities of warning. In other words, the practical possibility of being able to give a warning is not the determining factor in deciding whether they must give a warning.

The qualified nature of the warning rule reflects the historical weight attached to the importance of the element of surprise in military operations and the recognition that a commander might be justified in not providing a warning before an attack where the element of surprise is a necessary condition of its success (ICRC Commentary, para. 2223). Hence, while it may be feasible for an infantry team to issue a warning before approaching the command headquarters discussed previously, the on-scene commander is under no obligation to give advance warning if doing so would forewarn the enemy, thereby exposing the attacking force to ambush.

Another situation in which operational circumstances may preclude warning of an attack is where the speed of response by the attacking force is critical. This may be the case if an infiltrating ground force tasked with a specific mission is reallocated to conduct another attack at short notice because a fleeting opportunity exists to target an enemy commander whose location it is normally impossible to discern. When re-tasked at short notice in this way to target a time-sensitive target, the on-scene commander may have no access to the resources they might call upon in advance of a planned attack, such as leaflets or communication systems to broadcast warnings. They might instead be able only to give verbal or visual signals to the civilians who find themselves near an impending attack.

Ahead of an attack to clear a residential building the enemy is using, a commander might also choose not to give a warning encouraging civilians to leave that building if to do so would drive civilians into the already congested access routes the infiltrating force must use to conduct their attack such as corridors, stairs, and the streets surrounding the buildings. Such an operating reality would constrain the ability of the attacking force to manoeuvre towards and within the building, thereby negating any operational freedom gained by removing civilians from the localised point of attack.

Finally, as the purpose of an advance warning is to enable the protection of the civilian population, it is reasonable to expect that an attacking force does not give a warning ahead of an attack where the warning would increase the likelihood of civilian harm. This will be the case, for instance, if there is reliable intelligence to indicate the adversary will prevent civilians from evacuating or deliberately force civilians to serve as involuntary human shields after the attacking force has given notice of the timing and proposed location of an attack thereby increasing rather than mitigating the risk to civilians.

An adversary may alternatively use the warning to establish defensive positions along the infiltrating route with the risks of close-quarter combat in the streets presenting a greater danger to the civilian population than a surprise strike against the target originally identified. Yet due diligence is necessary in these cases because the exception is susceptible to misapplication by an attacking force, either as a pretext for foregoing the obligation to warn or as a means of unduly shifting responsibility for incidental loss to the enemy. Indeed, the benefits of a warning may even outweigh the risk presented to a potential group of human shields if it enables the evacuation of a greater number of civilians from the surrounding areas.


AP I, Article 57(2)(c) prescribes the giving of effective advance warning of attacks likely to affect the civilian population unless circumstances do not permit. In many battlefield situations this will be a straightforward assessment because military headquarters plan deliberate attacks ahead of time, and decision-makers have time to consider the practicalities of giving a warning. However, the complexities posed to ground troops by the urban environment often make it more difficult to determine the start and end points of an attack, or even whether the combat engagement in which they find themselves is of their own making such that the obligation to warn arises.

Even where it is possible to distinguish individual attacks, the operational circumstances make it more difficult to give effective warnings to all civilians likely to be affected by an attack. While this does not excuse soldiers from making an assessment, in many cases it means that soldiers may only be able to give verbal or visual signs to warn civilians in their immediate vicinity, who will often be aware of the impending military operations in any event. It concurrently means that soldiers will have no formal means of recording any warning given should they subsequently need to explain their actions.

These factors lead to the conclusion that a general warning given before ground forces enter the city is often the best, and sometimes only, means of warning all those civilians at risk from individual attacks that take place within the context of an overarching operation to take control of a city. It is the most an attacking force can do to mitigate the risks presented to the civilian population by combat action in the urban environment.


Lt Col Jon Griffiths is a legal officer with the British Army Legal Services. He is currently a legal adviser (operations) in the Office of Legal Affairs at NATO SHAPE. The views expressed are those of the author alone and not those of the British Army, the UK Ministry of Defence nor NATO.




Photo credit: IDF Spokeperson’s Unit