Following Israel’s 12-day air campaign in Iran in June 2025, the casualty figures reported by media sources, official reports, and humanitarian organisations varied. Why? Narges Qadirli explains how short urban air warfare exposes structural constraints in the recording, measurement, and verification of civilian harm across conflict datasets and casualty reporting systems
'How many people were killed?' During Israel’s June 2025 airstrikes on Iran, this question was the most frequently raised in public debate and often the hardest to answer. That is not because the harm did not happen, but because recording systems do not always capture the full extent of the harm civilians experience. For instance, conflict datasets and casualty reporting systems reveal measurable patterns of violence. Yet they do not fully reflect civilian harm in brief, intense urban air warfare.
Seven months after Israel’s air campaign, in January 2026, nationwide protests in Iran were met with state violence that caused mass civilian casualties. Weeks later, in February 2026, the US-Israeli war on Iran and the airstrikes that followed added another layer of violence and harm for civilians. When violent episodes happen back-to-back, documentation fragments and focus shifts. This makes assessments and reporting harder, and civilian suffering less visible, even if its effects continue.
Much of what we know about the June 2025 Israeli airstrikes comes from media coverage, non-governmental organisations such as Airwars, and event-based conflict datasets like ACLED. These systems operate differently.
Airwars investigates and assesses allegations of civilian harm associated with specific strikes through open-source verification. ACLED codes political violence events by time, place, and actor within a standardised framework. Both make air campaigns traceable. They show when violence peaks and where it concentrates. But the same logic that makes violence measurable can also narrow what counts as harm.
To analyse the events and fatalities in Iran during Israel’s airstrikes, I used ACLED’s Explorer for Iran with a focus on foreign military engagement in 2025. One of ACLED’s key indicators is 'exposure', which estimates how many people live in areas affected by violence.
As of 11 March 2026, the data show about 529 events, 1,079 fatalities, and 24.9 million exposure estimates. Tehran alone accounts for around 143 events. These figures cover all foreign military events in Iran in 2025, not just the Israeli airstrikes. However, June 2025, which coincides with the twelve days of the Israeli air campaign, stands out as the most violent month, with peaks in engagements, fatalities, and exposure.
Exposure to harm does not show who was hurt, who fled, who died, or who witnessed an event – only how many people lived under threat
Yet the data cannot provide a complete account of civilian impact. A closer look at ACLED records from mid-June in Tehran’s densely populated districts reveals uneven casualty reporting. Airstrikes on 15 and 24 June in Tehran’s District 5 exposed 752,050 people to harm. No fatalities were reported. By contrast, two strikes, on 13 and 23 June in northern District 1, with about 456,450 people exposed, caused around 62 and 80 fatalities respectively. Together, they account for roughly 174 of the fatalities recorded in District 1 during those twelve days.
These figures may change as new information becomes available. But the point is not only the numbers. It is how short air campaigns affect civilians in ways that remain obscured within quantitative assessments. I therefore read the data not only for what they record, but also for what they leave unclear. Exposure is not a measure of injury or loss. It does not show who was hurt, who fled, who died, or who witnessed an event. It estimates only the number of people living under threat.
From a qualitative perspective, what distinguishes these cases is legibility rather than geography or timing. ACLED’s 2025 data in Iran show widespread exposure alongside uneven confirmed fatalities. That combination produces only a partial representation of harm. Event-based datasets record incidents most confidently when deaths are reported and linked to a specific time and place. They cannot show how civilians experienced strikes and what unfolded inside homes, streets, and hospitals. Therefore, some harms become visible, while others remain delayed or absent from the record.
Event-based datasets cannot show how civilians experienced airstrikes, nor what unfolded inside homes, streets, and hospitals
In a city like Tehran, with roughly ten million residents, exposure estimates can be extremely high even for a single airstrike. When exposure appears alongside low or uncertain fatality counts, this does not mean civilians were unharmed. It means harm was not immediately measurable. Exposure signals the scale of risk, not the scale of suffering.
Different monitoring and recording systems analyse conflicts from distinct perspectives. For instance, NGOs like Airwars focus on investigating and verifying allegations of civilian harm at the case level using open-source methods. In contrast, event-based conflict datasets such as ACLED document political violence incidents more broadly, coding events and fatalities across different contexts within a consistent framework.
The main differences between such systems lie in their scope of analysis and methodology. However, they do not provide a complete picture and may face challenges during rapid, intense operations over short periods. During the 2025 Israel-Iran confrontation, limited access, fragmented reporting, and political sensitivity constrained casualty documentation across different systems. In these circumstances, the issue is not that the data are wrong, but that short, intense urban air campaigns strain the conditions under which civilian harm becomes measurable. As a result, reporting processes become more difficult, verification is delayed, and many forms of harm remain hard to capture because they develop gradually and indirectly and do not fit easily within these structures.
When multiple violent events occur within a short period, this may delay casualty verification. That does not mean the data is wrong
The gaps in documenting casualties from Israel’s airstrikes on Iran highlight the complexity of civilian harm assessments. Injuries can appear later, worsen, or lead to death. Infrastructure damage disrupts healthcare and livelihoods. Displacement, economic hardship, and trauma increase over time. These experiences alter civilian lives, but they hardly correspond to a single recordable event; eventually, they complicate the assessment of civilian harm.
While civilian harm assessments and event-based conflict datasets are essential for analysing patterns of violence and harm in aerial warfare, quantitative indicators alone are insufficient to show the true extent of civilian impact. In dense urban air warfare, many effects are too rapid, indirect, dynamic, or complex to verify. Without systematic qualitative assessment and on-the-ground documentation, important dimensions of civilian suffering remain unrecorded.
