Yes. A BBC investigation published in 2026 compiled 168 cases of Palestinian children shot in Gaza, 95 of them with wounds to the head or chest, and singled out two incidents, involving a 2-year-old and a 6-year-old, that occurred in places described at the time as a safe area and a safe corridor. That compilation is a news investigation built on testimony and medical records, not an official casualty register, and no ministry, statistical office or court record publishes a count of children shot disaggregated by whether the location carried a safe-zone designation. The phrase itself carries three different meanings across the accounts: a category the Fourth Geneva Convention establishes by agreement between parties, an area named in a unilateral Israeli evacuation instruction, and a descriptive label applied by news organizations and aid groups.
Key Findings
- A BBC investigation published in 2026 compiled 168 cases of Palestinian children shot in Gaza and reported that 95 of them were shot in the head or chest, with two of the cases occurring in areas described at the time as a safe area and a safe corridor.
- NBC News reported in 2025 that at least 7 minors were among 13 people killed while gathering water in al-Mawasi, an area the Israeli military had designated that day as a place for Palestinians to seek refuge. That incident is reported as a strike, not a shooting.
- Under Article 14 of the Fourth Geneva Convention of 1949, safety zones are established by agreement between the parties concerned. The Gaza areas at issue were designated unilaterally, by evacuation instruction.
- Israel’s Military Advocate General’s Corps reported in 2024 that it had opened 74 criminal investigations into alleged misconduct in the Gaza war and referred over 1,000 incidents to the General Staff’s Fact-Finding and Assessment Mechanism. Those are a party’s own counts of its own processes, and they are counts of investigations opened, not of findings.
- Becker, Ascone and Troschke (2022, peer-reviewed) coded 1,504 comments under leading British media Facebook pages during the May 2021 escalation and found 26.9% antisemitic; within those, 8.1% invoked child murder or blood libel.
Were children in Gaza shot in areas designated as safe?
Yes. A BBC investigation published in 2026 compiled 168 cases of Palestinian children shot in Gaza, and singled out two of them, a 2-year-old and a 6-year-old, in areas described at the time as a safe area and a safe corridor. The compilation is a news investigation assembled from testimony and medical records rather than an official casualty register.
Within that set, the BBC reported that 95 children were shot in the head or chest. That is a wound-location breakdown, drawn from clinical records and witness accounts case by case, and it is the most specific published quantity attached to the question.
Two properties of the figure govern how it can be read. It is a compilation, meaning it counts the cases an investigation was able to assemble and verify, so it functions as a floor on the number of such cases rather than an estimate of all of them. And it carries no denominator: the investigation reports how many cases it documented, not what share of children shot in Gaza those cases represent, because no published total of children shot exists against which to divide.
The safe-area description in the two singled-out incidents originates with the designations issued during the war. Israeli evacuation instructions named areas by number and directed civilians toward them; news organizations and aid organizations then used the resulting labels when describing where an event took place. The BBC’s account applies the description to the location at the time of the incident.
Is it often said no area in Gaza was consistently safe?
It is often said that no part of Gaza was consistently safe for civilians. The checkable form of that claim is movement, and UNICEF’s 2026 humanitarian situation update reports 1.9 million of Gaza’s 2.1 million people displaced. UNICEF footnotes the figure as a UN estimate from 2025 and states that many were displaced several times, so it is cumulative rather than a count of people displaced at any one moment.
The same document’s other headline figures carry weaker bases, and the difference matters for anyone using them. Its statement that nearly 90% of water and sanitation infrastructure has been damaged or destroyed appears with no source, no date and no denominator anywhere in the document, so what the share is a share of cannot be established from it. Its statement that 82% of families are water insecure originates with the WASH Cluster’s Light Touch Monitoring in Gaza in March 2026, a phone survey of 342 assessed households whose authors state the findings cannot be considered representative of the population in Gaza.
At population scale, two records are firmer. The Palestinian Central Bureau of Statistics reported in 2025 that Gaza stood at about 2.13 million people at the end of that year, a decline of approximately 254,000, which the Bureau states is a decrease of 10.6% against its pre-aggression population estimates. The Bureau separately reports that roughly 2.2 million people were living in Gaza on the eve of the war and that nearly two million of them were displaced. The Bureau attributes the change to war-related deaths, missing persons, reduced births and forced out-migration; it is a demographic net decline, not a casualty count.
On the physical record, the Rapid Damage and Needs Assessment published jointly in 2026 by the World Bank, the European Union and the United Nations reports more than 371,888 housing units destroyed or damaged and more than 50% of hospitals non-functional, from remote sensing combined with ground-verified data.
These aggregates document displacement and destruction across the territory. None of them resolves what happened at any single designated location, which is the question the compiled cases address.
What does the phrase safe zone actually designate?
Under Article 14 of the Fourth Geneva Convention of 1949, safety zones are established by agreement: the text provides for zones organized to protect the wounded, the sick, the aged, children under fifteen, expectant mothers and mothers of children under seven, and for agreements between the parties concerned on their recognition. The areas at issue in Gaza were designated unilaterally.
Israel’s own published materials describe the mechanism it used. The Israel Defense Forces states that it warned civilians by telephone calls, text messages and leaflets, used a loud non-lethal munition against a building’s roof before striking it, aborted strikes when civilians were present, and published a numbered evacuation-area map so that residents could follow instructions by area. The IDF publishes no counts for any of this: no number of warnings issued, roof-knocks performed or strikes aborted appears in the materials, so the measures can be described but not measured.
Media usage follows the designation rather than the treaty. NBC News reported in 2025 on an incident in al-Mawasi, describing the area as one the Israeli military had designated that day as a place for Palestinians to seek refuge. The wording attaches the status to a date and an instruction, not to a standing recognized zone.
Three distinct objects therefore travel under one phrase. There is the treaty category, which requires agreement between belligerents and a recognized boundary. There is the operational designation, issued by one party as part of an evacuation instruction and revisable by that party. And there is the descriptive label, applied afterward by news organizations and aid organizations to areas where displaced civilians had gathered, sometimes rendered as a humanitarian zone or a safe corridor.
A reader comparing two accounts of the same location is often comparing two of these three. The consequence is practical: a statement that an incident occurred in a safe zone can be accurate in the second or third sense while establishing nothing about the first, and the accounts rarely say which sense is in use.
How many children were shot rather than killed by strikes?
Of the 168 cases the BBC compiled in 2026, 95 children were shot in the head or chest. That wound-site breakdown exists because gunshot cases are recorded one by one from clinical and witness evidence; strike casualties are counted by a different method and produce no comparable figure.
The clinical observation is corroborated by named reporting from earlier in the war. The Guardian reported in April 2024 that physicians returning from Gaza described children arriving with single gunshot wounds to the head and chest. That is testimony from practitioners about what they saw in the facilities where they worked, and it is not a sample: the physicians report a pattern in the patients who reached them, which cannot be converted into a rate.
The two phenomena must be held apart. A gunshot case compilation counts identified individuals and records where each was wounded. A strike casualty figure counts deaths from an explosive event, is usually produced by a health ministry or a news organization at the scene, and carries no wound-site breakdown. The compiled shooting cases are a far smaller number than the aggregate casualty totals each body publishes for the war, and the two cannot be summed, because they are produced by different means on different definitions and overlap is not documented either way.
That distinction is what a phrase such as “children were killed in safe zones” tends to collapse. Reporting that gathers strike deaths, gunshot cases and general casualty totals under one sentence produces a claim no single record supports on its own. The reporting on civilian harm in Gaza contains all three kinds of record, and each answers a different question.
No published figure states what share of children killed or wounded in Gaza were shot rather than struck. That share is a quantity no register in the public record is built to produce, because the ministry totals are not disaggregated by weapon and the case compilations have no denominator.
What is documented about children killed in al-Mawasi?
NBC News reported in 2025 that at least 7 minors were among 13 people killed while gathering water in al-Mawasi, an area the Israeli military had designated that day as a place for Palestinians to seek refuge. That incident is reported as a strike rather than a shooting, and it is the most specific figure attached to a named designated area in the published accounts.
Al Jazeera reported in 2025 that at least 5 children were killed in a strike on tents in al-Mawasi, an area the outlet described as a humanitarian safe zone. That is the outlet’s own account and its own characterisation of the area’s status, and the outlet holds a stated position on the conflict.
What the al-Mawasi record consists of, then, is a set of individually reported strike incidents, published by news organizations at or near the time, each carrying its own casualty figure and its own description of the location. There is no consolidated tally: no body publishes a count of incidents in al-Mawasi, a count of children killed there, or a breakdown of those deaths by weapon type.
Israel’s own published materials do not close that gap. The Coordinator of Government Activities in the Territories publishes detailed supply figures for the humanitarian area during the ceasefire period, including 1,778,363 metric tons of recorded food entry between 10 October 2025 and 7 June 2026, on its own self-declared basis rather than weighing at the crossings. It publishes no incident-level record for the area. The Israel Defense Forces publishes evacuation-area maps and its account of the warning procedures used, and no counts of events within the designated areas.
The result is an asymmetry in the record. The volume of aid moved into the area is documented to the ton by the party that moved it, and the events that took place inside it are documented incident by incident, by outside news organizations, with no register on either side of the line.
What does Israel’s own record state about the humanitarian area?
Israel’s Military Advocate General’s Corps reported in 2024 that it had opened 74 criminal investigations into alleged misconduct in the Gaza war, and that over 1,000 incidents had been referred to the General Staff’s Fact-Finding and Assessment Mechanism. These are the Corps’ own counts of its own processes, published by a party to the events and not independently verified.
The 74 break down as 44 investigations concerning the death of detainees held by the IDF, 13 alleged pillaging or stealing, 8 alleged detainee mistreatment, 6 alleged illegal use of force, and 3 alleged destruction of civilian property without military necessity. None of the five categories is a shooting of a child in a designated area, and the published breakdown does not include one.
The threshold that produces those counts is stated by the Corps in its own words: in armed conflict, “death or injury to a civilian or harm to civilian objects, while tragic and regrettable, in most cases do not by themselves indicate a reasonable suspicion of criminal conduct, as they can be explained as reasonable aftermath of lawful operational activity.” The investigation count is a function of that standard, so the ratio between 74 and the referred incidents is uninterpretable without it. The counts are also of investigations opened, not of findings, indictments or convictions.
Alongside the machinery sits the stated requirement. The army’s own ethical code states, under the value of Purity of Arms, that a soldier “will not use their weapon or power to harm uninvolved civilians and prisoners and will do everything in their power to prevent harm to their lives, bodies, dignity and property.” The code records what the IDF requires of itself; it is evidence of the stated standard and not of conduct against it.
Can wound patterns establish that a child was targeted?
Wound site is recordable; the shooter is not. 95 of the BBC’s 168 compiled cases carry a head or chest wound location, and the 2026 investigation reports no chain-of-custody forensic examination establishing who fired in any of them. A wound record documents what a body sustained, and stops there.
Three separate propositions are in play whenever such a case is described, and they are established by different evidence. That a child was shot is established by clinical examination and, in these cases, by medical records the investigation obtained. Where the child was is established by witness accounts, location data and, in some cases, contemporaneous imagery. Who fired, and on what order, is established by none of the above.
The clinical testimony reported by The Guardian in April 2024, in which physicians described children arriving with single gunshot wounds to the head and chest, illustrates the boundary precisely. Wound location and the observation that a wound was single rather than one of several are recordable facts about a patient. Neither identifies a weapon, a position or a firing party, and the reporting does not connect the pattern to an identified shooter.
Nor do these methods produce a rate. A compilation counts the cases it assembled; a clinician reports what reached the hospital where they worked. Neither has a denominator, so neither supports a sentence of the form “X% of children shot were hit in the head”, however often the underlying figures are recombined into one.
Intent, where a legal process assesses it, is established on a different class of material entirely: orders, rules of engagement, targeting records, command communications and testimony from those who issued them. None of that appears in a medical file. The Military Advocate General’s Corps states the same boundary from the other direction when it writes that civilian death or injury in armed conflict does not by itself indicate reasonable suspicion of criminal conduct. The clinical record and the command record are not substitutes, and the published material on these cases is almost entirely the former.
Has any body published a count of these shootings specifically?
No official register disaggregates children shot in Gaza by whether the location carried a safe-zone designation. The 168 cases compiled by the BBC in 2026 remain the largest published set on the subject, and it is a news compilation rather than a casualty register. No ministry, statistical office, court record or commissioned review publishes the breakdown the question asks for.
What exists instead are three registers built on three other axes. The Palestinian Ministry of Health publishes a cumulative death total for Gaza, reported as 70,942 by the end of December 2025 and relayed by the Palestinian Central Bureau of Statistics, which states it did not measure it; that total is not disaggregated by age, weapon and location status together. Israel’s Military Advocate General’s Corps publishes counts of investigations opened, which record the state’s own processes rather than casualties. And news organizations publish incident-level accounts, which record single events with no common definition of what counts as a designated area.
The UN human rights office issued reporting in 2026 stating that Israeli security forces targeted children in shelters and designated safe zones. That is a stated finding by that body, on its own definitions, and it is published without a disaggregated count of children shot by location designation. UN reporting on the period since the October 2025 ceasefire counts child casualties overall rather than by weapon type or by the status of the place where each occurred.
The absence is itself a finding, and it explains why accounts of this question diverge so widely. A reader encountering a confident number is encountering either a case compilation, which is a floor with no denominator, or an aggregate casualty figure, which measures something broader than the question. No published source reconciles the two, and no body has assembled a register on the axis the question is asked along.
How does child-harm framing appear in measured discourse?
Becker, Ascone and Troschke (2022, peer-reviewed) coded 1,504 comments posted under leading British media Facebook pages during the May 2021 escalation and found 26.9% antisemitic; within that antisemitic set, 8.1% invoked child murder or blood libel.
The same study coded 1,500 French comments at 12.6% and 1,520 German comments at 13.6%.
The remaining British breakdown is dominated by other frames: an evil stereotype in 39.8% of the antisemitic comments, Israel’s sole guilt in 27.9%, denial of Jewish self-determination in 7.7%, an apartheid analogy in 5.2% and a Nazi analogy in 4.2%. The authors also report that thread context was required to infer the antisemitic meaning in 38.7% of the British antisemitic comments, so a keyword reading of the same corpus would have missed a substantial share.
Ozer, Obaidi and Bergh (2025, peer-reviewed) ran two experiments with British samples, of 354 and 490 participants, priming participants with war-in-Gaza material and measuring hostility toward Jews. They report two distinct pathways: political orientation associated with anti-Jewish hostility through social dominance orientation, and conservative orientation negatively associated with antisemitism through out-group identity fusion with the Palestinian people. Both are measured associations in an experimental setting, not claims about cause in the wider population.
Huesmann and colleagues (2012, peer-reviewed) surveyed 89 Jewish American and 180 Arab American high-school students on media exposure to the conflict and ethnic stereotyping. Total media exposure did not directly predict stereotypes; the association ran through identification with the groups as portrayed in the media.
Methodology and limitations
This report assembles published records on a single question and states what each record measures and how it was produced. It draws on four source types, labelled at each use: treaty text (the Fourth Geneva Convention of 1949, Article 14); official records published by a party to the events (the Israel Defense Forces, its Military Advocate General’s Corps, the Coordinator of Government Activities in the Territories, and the Palestinian Central Bureau of Statistics); peer-reviewed research (Becker, Ascone and Troschke, 2022; Ozer, Obaidi and Bergh, 2025; Huesmann and colleagues, 2012); and named news investigations (the BBC, 2026; NBC News, 2025; The Guardian, 2024).
Figures are carried as their publishing sources report them. Case compilations by news organizations are floors: they count what an investigation assembled and verified, carry no denominator, and are not casualty registers. Counts published by a party to the conflict are that party’s own reporting of its own processes or supplies, not independently verified, and the Military Advocate General’s investigation counts are counts of investigations opened rather than of findings, indictments or convictions. Aid-agency situation updates frequently relay figures measured by other bodies; where a relayed figure could be traced, it is attributed to the body that measured it, and where a figure appears without a source, date or denominator in the document carrying it, that is stated.
The central limitation is a documented absence. No register in the public record disaggregates children shot in Gaza by whether the location carried a safe-zone designation, and none reconciles gunshot case compilations with aggregate casualty totals produced by different methods. Statements about who fired, and on what order, rest on material that does not appear in clinical records or case compilations.
The final section measures discourse rather than events: it reports how frames appear in coded comment corpora and in survey experiments, in correlational terms and with each study’s sample stated. No individual is named or profiled anywhere in this report.
Conclusion
So were children in Gaza shot in areas designated as safe? The compiled record answers yes, and it answers in a specific shape: 168 cases assembled by a news investigation in 2026, 95 of them with head or chest wounds, two occurring in places described at the time as a safe area and a safe corridor. That is what the strongest published set contains, and it is a floor without a denominator.
Around that set, the records diverge by design. The phrase itself resolves three ways, into a treaty category the Fourth Geneva Convention builds on agreement between parties, an area named in a unilateral evacuation instruction, and a label applied afterward by outside observers. What is documented at al-Mawasi is chiefly strike casualties reported incident by incident, including at least 7 minors among 13 people killed gathering water in an area the Israeli military had designated for refuge that day. Israel’s own machinery published 74 criminal investigations against over 1,000 referred incidents, together with the threshold that produced both numbers, and 1,778,363 metric tons of food entry it recorded itself. And nowhere among the ministry totals, the state’s investigation counts and the news compilations is there a register built along the axis the question travels on.
The imagery does not stay inside the file it came from. Becker, Ascone and Troschke found that 8.1% of the antisemitic comments they coded under British media posts reached for child murder or blood libel, a frame centuries older than any evacuation map, sitting in the same threads as the reporting. That is the point at which a case record stops being a case record and becomes something else in circulation. What happens to an image of a harmed child on the way from a medical file to a comment about Jews, and how much of that route was already laid before the war began, is the part of this subject that has not been counted.
Frequently Asked Questions
Why do two accounts of the same location disagree about whether it was a safe zone?
Because the phrase is doing three jobs. Article 14 of the Fourth Geneva Convention of 1949 defines safety zones as established by agreement between the parties concerned; Israeli evacuation instructions designated areas unilaterally and by date; news organizations and aid groups applied descriptive labels afterward. An account can be accurate in one sense and silent on the others.
What is the difference between a compiled case count and a casualty register?
A compilation counts the individual cases an investigation could assemble and verify, so it functions as a floor and carries no denominator. A casualty register is maintained continuously by a health authority or statistical office and reports totals. The BBC’s 168 cases are the former; the death totals attributed to the Palestinian Ministry of Health are the latter, and they are not comparable quantities.
How does the Fourth Geneva Convention treat children specifically?
Article 14 names children under fifteen, along with the wounded, the sick, the aged, expectant mothers and mothers of children under seven, as the categories that safety zones may be organized to protect. The provision is about establishing recognized zones by agreement, not about the conduct of hostilities generally.
Why does Israel’s investigation count look small beside the number of incidents referred?
Because the Military Advocate General’s Corps applies a stated threshold: in armed conflict, in its own words, civilian death or injury “in most cases do not by themselves indicate a reasonable suspicion of criminal conduct.” The 74 investigations are a function of that standard, and citing the count without the standard leaves a ratio a reader cannot interpret.
What has Israel published about how civilians were warned?
The Israel Defense Forces states that it used telephone calls, text messages, leaflets, a loud non-lethal munition on rooftops before strikes, and a numbered evacuation-area map. It publishes no count of warnings issued, roof-knocks performed or strikes aborted, so the measures are described in its materials without a magnitude.
Which measured research connects conflict coverage to attitudes toward Jews?
Ozer, Obaidi and Bergh (2025) ran two British experiments, of 354 and 490 participants, priming war-in-Gaza material and finding two distinct pathways to anti-Jewish hostility. Huesmann and colleagues (2012), surveying 269 Jewish American and Arab American students, found media exposure did not directly predict stereotyping and that the association ran through identification with the groups as portrayed.
How much of Gaza’s population moved during the war?
UNICEF’s 2026 humanitarian update reports 1.9 million of 2.1 million people displaced, relaying a 2025 UN estimate and noting many were displaced several times, which makes it cumulative rather than a point-in-time count. The Palestinian Central Bureau of Statistics separately reports roughly 2.2 million residents on the eve of the war and nearly two million of them displaced.
Sources
- BBC News, 2026. Investigation into Palestinian children shot in Gaza. bbc.com. News investigation.
- Becker, M., Ascone, L., and Troschke, H., 2022. Antisemitic Comments on Facebook Pages of Leading British, French, and German Media Outlets. Humanities and Social Sciences Communications. DOI 10.1057/s41599-022-01337-8. Peer-reviewed.
- Coordinator of Government Activities in the Territories (COGAT), 2026. Humanitarian Situation Report During the Ceasefire. gaza-aid-data.gov.il. Official record, a party’s own self-reported counts.
- Geneva Convention (IV) relative to the Protection of Civilian Persons in Time of War, 12 August 1949, Article 14. International Committee of the Red Cross treaty database, ihl-databases.icrc.org. Treaty text, primary record.
- Huesmann, L. R., Dubow, E. F., Boxer, P., Souweidane, V., and Ginges, J., 2012. Foreign Wars and Domestic Prejudice: How Media Exposure to the Israeli-Palestinian Conflict Predicts Ethnic Stereotyping by Jewish and Arab American Adolescents. Journal of Research on Adolescence. DOI 10.1111/j.1532-7795.2012.00785.x. Peer-reviewed.
- Israel Defense Forces. The Spirit of the IDF: the ethical code and its values. idf.il. Official record, a party’s own published doctrine.
- Israel Defense Forces. Stated precautions in Gaza and the human-shields position. idf.il. Official record, a party’s own stated position.
- Israel Defense Forces, Military Advocate General’s Corps, 2024. Addressing Alleged Misconduct in the Context of the War in Gaza. idf.il. Official record, a party’s own reported counts.
- NBC News, 2025. Reporting on casualties in al-Mawasi. nbcnews.com. News reporting.
- Ozer, S., Obaidi, M., and Bergh, R., 2025. The Impact of Globalized Conflicts: Examining Attitudes Toward Jews Among Britons in the Political Context of the War in Gaza. International Journal of Intercultural Relations, vol. 107. DOI 10.1016/j.ijintrel.2025.102184. Peer-reviewed.
- Palestinian Central Bureau of Statistics, 2025. A Brief on the Status of the Palestinian People at the End of 2025. pcbs.gov.ps. Official record.
- The Guardian, 2024. “Not a normal war”: doctors say children have been targeted by Israeli snipers in Gaza. theguardian.com, 2 April 2024. News reporting.
- UNICEF State of Palestine, 2026. Humanitarian Situation Update: Analysis of the Response to the Needs of Children at Mid-year 2026. unicef.org/sop, 17 July 2026. Agency situation update and funding appeal.
- WASH Cluster, 2026. Light Touch Monitoring, Gaza, March 2026. Cluster monitoring survey of 342 assessed households, stated by its authors as not representative.
- World Bank, European Union and United Nations, 2026. Gaza Strip Rapid Damage and Needs Assessment. thedocs.worldbank.org. Joint intergovernmental assessment.