Child road casualties rise sharply in Northern Ireland as 115 are killed or seriously injured

More children were killed or seriously injured on Northern Ireland’s roads in 2025 than in the previous year, with the official total rising by 22. The increase is a clear warning sign, but the headline figure cannot by itself explain where the incidents happened or what caused them.

What the figures show

There were 115 children under 16 killed or seriously injured in police-recorded road traffic collisions in Northern Ireland during 2025. The equivalent figure for 2024 was 93. That is an increase of 22 children, or about 24% year on year.

The figures were published by statisticians working with the Police Service of Northern Ireland (PSNI) and Northern Ireland Statistics and Research Agency (NISRA) in the annual calendar-year road-collision statistics. Across all ages, 56 people died and 1,066 were seriously injured on Northern Ireland roads in 2025. The combined total of people killed or seriously injured reached 1,122, up 11.3% on 2024.

The child figure is therefore not an isolated measurement of every kind of road injury. It focuses on children who died or sustained serious injuries, rather than all children recorded with slight injuries. That distinction matters: a change in serious outcomes should not be diluted by a larger but less severe category.

A rise that needs explaining—not guessing

The annual increase tells us that the number of children experiencing the most serious recorded outcomes rose. It does not identify the reason. The figures do not, on their own, show whether the change was associated with pedestrian collisions, children travelling as passengers, cycling, particular road layouts or a small number of incidents involving several casualties.

It would be premature to blame speeding, road design, driving behaviour or any particular area without checking the collision-level records. Serious child casualties can occur in very different circumstances, and a national total can conceal meaningful differences between those circumstances.

The next stage of a rigorous investigation is to link child-casualty records to the corresponding collision records. That allows each child’s recorded age group, road-user type, injury severity and the collision’s available location and circumstances to be examined together. Records with missing location or classification information must be identified rather than silently treated as if they were complete.

Why the way children use roads matters

Children are not one uniform group of road users. A child walking to school encounters a different environment from a child travelling as a passenger on a rural road. Older children may also cycle or travel independently in situations that younger children do not.

For that reason, a useful analysis would separate pedestrians, cyclists, vehicle passengers and any other road-user categories available in the underlying records. It should then compare serious injuries and deaths separately where sample sizes permit. If a pattern appears in one category, that finding should be checked against the number of records with usable information and against earlier years using consistent definitions.

Age-specific results also need careful wording. An increase in the number of child casualties is not, by itself, proof that an individual child’s journey became more dangerous. That conclusion would require suitable information about exposure—such as journey numbers or distance travelled—and a fair comparison across periods.

What PureCar’s records can add

PureCar has assembled a historical archive of Northern Ireland collision, casualty and vehicle records. The value of that archive is not simply repeating a published headline; it is providing a basis for checking patterns across years and asking more precise questions of the underlying records.

For this story, the next analysis should test whether the 2025 rise is spread across road-user types and locations or concentrated in particular circumstances. Any geographic comparison must use a consistent definition of location and a verified join between casualty and collision records. A road with more recorded incidents is not automatically more dangerous: traffic volumes, journey exposure and the road’s length can differ substantially.

The analysis should also reconcile totals with PSNI’s published tables, document missing fields and avoid treating multiple child casualties from one collision as multiple collisions. Those checks are essential before presenting a map, ranking or apparent cluster as a finding.

What families and drivers can take from this

The figures do not point to one single remedy, but they reinforce the importance of protecting children in and around traffic. Drivers should be especially alert around schools, residential streets and places where children may cross unexpectedly; choose a speed that allows them to stop within the visible clear distance; and avoid distraction. Children travelling in vehicles should use an appropriate restraint for their size and age.

Those are general safety measures, not conclusions that any one behaviour caused the reported increase. The evidence-led public question is more specific: what do the collision records reveal about the circumstances behind the additional 22 children, and are the same patterns visible over a longer period?

What happens next

PureCar’s next step is to compare the child casualty records for 2024 and 2025 with the historical series, then examine available road-user and collision-context fields. Findings should only be published after the records have been validated and the counting unit is explicit.

The key result is unambiguous: 115 children under 16 were killed or seriously injured in 2025, up from 93 in 2024. That increase deserves attention. Explaining it responsibly means going beyond the headline and letting the detailed records—not assumptions—determine the next story.

How to tell whether the change is widespread

A year-on-year increase can come from several different patterns. The number may rise across multiple road-user groups, or a smaller number of collision types may account for much of the difference. The yearly total alone cannot distinguish those possibilities. A responsible follow-up should show both the number of affected children and the number of collisions in which they were involved, because the two measures answer different questions.

Analysts should compare like with like: calendar year against calendar year, the same age definition, the same severity categories and the same geographic coverage. If a record has no reliable age or road-user classification, it should be counted as unclassified and included in the limitations. Quietly excluding incomplete records can make a result look more certain than it is.

What a local analysis should include

For each location grouping, a useful report would state the number of child casualties, the number killed or seriously injured, the total number of relevant collisions and the proportion of records with usable location data. It should make clear whether a location means a council area, a road segment, a junction or a broad settlement category. Mixing those definitions would make comparisons unreliable.

Small local totals should be interpreted cautiously. A change from one child casualty to three is an increase of 200% mathematically, but the small numbers do not prove a persistent local trend. Multi-year comparisons, transparent counts and context about the underlying road network are more informative than rankings designed only to produce a dramatic headline.

PureCar should also distinguish an observed pattern from a recommendation. If several serious child casualties appear near a particular road feature, that may justify further investigation; it does not by itself prove that the feature caused the collisions or that a particular engineering change would prevent them. The next step would be to examine the site and seek supporting evidence from the responsible road authority.

Keeping the reporting accountable

The strongest public-interest version of this story will publish the calculation alongside the caveats. Readers should be able to see the base figures, the comparison period and how the percentage was calculated. If PureCar later identifies a local pattern, the article should describe how the records were linked and how many cases were omitted because their information could not be verified.

That approach is less sensational than assigning blame from a national figure, but it is more useful to families, policymakers and motorists. The 22-child increase is enough to warrant attention on its own. The follow-up should show what the detailed evidence says, while being equally clear where it cannot yet provide an answer.

Methodology: This report uses the published PSNI/NISRA calendar-year totals for 2024 and 2025. The year-on-year difference and percentage change are calculated from those published totals. No unverified road-level hotspot or cause is inferred from the aggregate figures.