NEWS 05 November 2026

Northern Ireland's road-safety paradox: deaths fell, serious injuries rose—and the child casualty figure is worse

Published 5 November 2026 at 23:29

NI NEWS

The official 2025 figures look reassuring if judged only by fatalities. Combining PSNI collision data, strategy indicators and information disclosures reveals rising severe harm, a widening gap against key baselines and significant holes in what local collision records capture.

A year in which fewer people were killed on Northern Ireland's roads ought to be welcome. But the full account of 2025 reveals why a single headline number can give a dangerously incomplete impression. Fifty-six people died, compared with 69 in 2024. Over the same period, the number of people seriously injured rose from 939 to 1,066. The combined count of people killed or seriously injured—known as KSI—rose from 1,008 to 1,122. In other words, deaths fell by 13, while the number of people experiencing the most severe recorded outcomes increased by 114.

The picture is more troubling for children. The official count of under-16s killed or seriously injured rose from 93 to 115, an increase of 22 children, or approximately 24%. Against the road-safety strategy's baseline of 71, the 2025 figure is around 62% higher. Those figures come from the Police Service of Northern Ireland and Department for Infrastructure's statistical releases, not from a collection of social media posts or isolated crash reports. They show why any claim that road safety is improving must answer a harder question: improving for whom, and against which measure?

The outcome that dominates the headlines

The PSNI's calendar-year summary for 2025 recorded 5,015 injury collisions causing 7,879 casualties. In 2024 there had been 4,753 injury collisions and 7,494 casualties. So while the number of deaths moved in the right direction, injury collisions and total casualties increased. The rise in serious injuries was particularly pronounced: 127 more people were seriously hurt in 2025 than in the previous year. These are changes between calendar years, not financial years, and they should not be blended with separate April-to-March accident tables.

Fatalities are a vital public-health indicator, but they are also a comparatively small number that can fluctuate considerably between years. One collision can kill several people; another can leave people with severe, lifelong injuries but no fatality. Annual death totals can therefore improve while the wider burden of serious injury gets worse. That is not an argument for discounting the reduction in deaths. It is an argument for measuring both results together rather than allowing one to obscure the other.

The relationship is visible in the official KSI series. The 2025 total of 1,122 was around 11.3% above 2024's 1,008. That is not a rounding error or a disagreement over terminology; it is the combined effect of fewer fatalities and many more serious injuries. For somebody recovering from a collision, the difference between fatal and serious harm is obviously profound. For the health service, emergency responders, families and employers, the increase in severe injuries is also a major social outcome that a death-only narrative misses.

The longer trend contradicts the easy narrative

The PSNI's published history provides context. In 2019, Northern Ireland recorded 56 road deaths and 774 serious injuries, making 830 KSIs. The 2025 death total was also 56, but the serious-injury count was 1,066, and the KSI total 1,122. The comparison is striking: identical annual fatalities, yet 292 additional people killed or seriously injured in 2025 compared with 2019, driven by the increase in serious injuries.

This does not prove every element of danger on the roads worsened by the same amount. Exposure can change: miles travelled, vehicle use, road-user mix, reporting practice and the distribution of severity may vary. But the point of a serious statistical investigation is not to dismiss the recorded rise simply because several explanations might exist. It is to identify the rise accurately and then ask which factors, or combination of factors, account for it.

The 2020 figures show why historical comparisons also need care. That year recorded 56 deaths but 596 serious injuries, during a period when travel patterns were disrupted by the pandemic. Treating 2020 as a normal baseline would flatter or exaggerate subsequent trends in ways that have little to do with road-engineering decisions. That is one reason the road-safety strategy uses multi-year baseline averages rather than selecting a convenient single year.

What the 2030 strategy says about success

DfI's 2025 performance report, published in August 2026, compares results with baselines based generally on 2014–18 averages. It reports a baseline of 68 road fatalities, 751 seriously injured people, 71 children killed or seriously injured and 196 KSIs among people aged 16–24. Against those points of reference, the 2025 fatalities figure was 17% lower. Serious injuries, however, were 42% above baseline; child KSI was 62% higher; and young-people KSI was 16% higher.

Those are four different measures, and they tell four different stories. Progress on deaths is genuine within that comparison. It is not evidence of comparable progress on the other three. When public bodies speak about meeting road-safety goals, the balance matters: an outcome framework exists precisely to prevent one favourable trend from hiding unfavourable ones. A policy that reduces fatal consequences but leaves serious injuries rising cannot be assessed solely by how many deaths it prevented.

The strategy also uses rolling averages for some indicators, reducing the influence of a particularly bad or good single year. That is methodologically sensible. But it does not erase the need to explain a sustained gap between serious-injury outcomes and the target baseline. The question is whether the interventions being implemented are addressing the types of collision, road users and locations driving the higher burden.

The child casualty rise deserves its own investigation

The movement from 93 child KSIs in 2024 to 115 in 2025 is the sharpest warning in the summary. The figures group together fatalities and serious injuries among children under 16; they do not mean 115 children were killed. Failing to make that distinction would cause needless alarm and weaken the reporting. Yet the combined harm is serious enough without exaggeration. Twenty-two additional children experienced outcomes in the most severe police-recorded category in a single year.

The PSNI's earlier financial-year reporting for 2024–25 showed that nearly three-quarters of all injured children were vehicle passengers, while roughly one-third of children killed or seriously injured were pedestrians. Those refer to a different reporting period and different severity populations, so they should not be used as a detailed category breakdown for the 115 calendar-year cases in 2025. They do, however, demonstrate why child road safety cannot be reduced to whether children are using crossings properly: children are exposed as passengers, pedestrians and other road users.

For the new 2025 cohort, the obvious questions are how many cases involved school travel, built-up streets, rural roads and vehicle occupants; whether there are recurring junction or speed-limit characteristics; and what injuries occurred by age band. The headline series does not by itself provide all of those answers. A report should obtain linked, carefully anonymised tables before drawing conclusions about the specific environments in which the rise occurred.

A young-driver and young-road-user problem—but not only theirs

For 16–24-year-olds, the number killed or seriously injured rose from 207 to 227 between 2024 and 2025, according to the strategy comparison. That is an increase of 20, or around 10%. The 2025 group accounted for approximately one-fifth of all KSIs. This age group attracts attention because of inexperience and known road-safety risks, but the figures refer to young people killed or seriously injured, not solely young drivers causing collisions.

That distinction prevents a familiar distortion: treating every young casualty as evidence of fault by a young motorist. A 19-year-old passenger, a young pedestrian and a novice driver are all within the age total, but the underlying intervention may be different in each case. Driver training could matter to one set of events; street design, safe speed and occupancy conditions might matter to others. An age-based KSI count does not identify responsibility.

A separate NISRA analysis of speed-related collisions covering 2019–23 found that 41% of people killed or seriously injured in speed-related collisions were aged 16–24, and that approximately 70% of speed-related KSI collisions occurred on rural roads. This is a different five-year analysis from the 2025 calendar-year report. It strengthens the case for examining age, rural exposure and speed together, but cannot be read as saying 41% of all 2025 KSIs were due to young people's speeding.

Rural roads and the difference between count and rate

The road-safety strategy reports that 43 of the 56 people killed in 2025 died in collisions on rural roads. That is approximately 77% of the annual road-death total. It is an important distribution and supports focused examination of rural-road risks. It is not, by itself, a calculated risk rate. To say rural roads were a certain number of times more dangerous would require consistent measures of vehicle kilometres, trip patterns and exposure for the relevant road categories.

Official figures also say Mid Ulster recorded nine road deaths in 2025, while Derry City and Strabane recorded none, and Belfast had 180 KSI casualties. The contrast is politically and emotionally charged, but it needs context. A district with a very small number of deaths in one year can see a large percentage change after a handful of collisions. Belfast's population, travel demand and non-resident traffic differ from those of rural districts. Ranking local authority safety solely by raw deaths or KSIs would risk overstating the significance of arbitrary boundaries and short periods.

For targeted policy, a five-year exposure-adjusted series would be more useful than one year's local ranking. Road-user category, time of day, road class and actual collision location should be part of the calculation. It is entirely possible for an urban area to have many injuries but fewer deaths and for a rural area to have fewer collisions with a higher proportion of severe outcomes. The current annual summary alone cannot confirm which mechanism explains each district's position.

The striking difference between forms of travel

One of the strongest comparisons in DfI's 2025 strategy release concerns KSIs per distance travelled. It reports approximately four KSIs per 100 million kilometres travelled by car or van, compared with 344 per 100 million kilometres travelled by motorcycle. The ratio of the two published rates is 86 to one. Pedal cyclists and pedestrians are also shown with higher rates than car users—about 63 and 41 per 100 million kilometres, respectively, in their corresponding travel measures.

Those mode comparisons are consequential, but they should be explained carefully. They are aggregate population measures using exposure estimates, not a forecast that any particular motorcycle journey is 86 times more likely to end in a serious crash than a particular car trip. Trip lengths, journey types, road environments, age and conditions all influence exposure. The source method and confidence around kilometres travelled matter. Even with those qualifications, the huge difference in reported rates is a compelling reason not to design road safety exclusively around protecting people inside cars.

A person on a motorcycle or bicycle has different vulnerability to road-surface defects, collision forces and vehicle movements. A pedestrian has no protective structure. The need for safe street layouts, enforcement and road maintenance follows from the scale of potential harm, not from a belief that all users face equal risk. The strategy's mode-specific measurements help move the debate away from treating every casualty as interchangeable.

Causation figures must not be turned into simplistic blame

The PSNI's 2025 analysis lists inattention or diverted attention as the principal causation factor associated with 115 KSI casualties, followed by wrong course or position with 95, and excessive speed having regard to conditions with 86. These are police-recorded assessments associated with serious outcomes. They are valuable for identifying recurring factors in collisions, but they are not a complete scientific allocation of fault and do not mean there were exactly 115 collisions involving distraction.

The Road Safety Partnership's summary separately says excess speed was a factor in 11 of 53 fatal collisions in 2025. Seven were recorded with speed as a principal reason and four as a contributory factor. The difference between fatal collisions (53) and people killed (56) is not an inconsistency: a single collision can result in more than one death. It is an example of why collision counts and casualty counts must remain separate in analysis.

Police coding may identify a principal factor while other contributing circumstances also matter. A road design, poor visibility, driver behaviour, traffic speed and vehicle condition can interact. A documentary investigation can legitimately test whether authorities are concentrating interventions on the factors most commonly present. It cannot infer the details of individual cases from a national table.

What the police do not count centrally

A further limitation appears in published PSNI information responses about specific roads. In FOI/16713, concerning Crosshill Road near Crumlin, the police said the injury-collision statistics were available through their statistical system but damage-only road traffic collisions were not held in a central place that could be searched in the same way. In another site-specific response about the Glenravel Road, PSNI explained that identifying damage-only crashes would require manual examination of records and could exceed FOI cost limits.

That means a publicly cited count of 'collisions' on a named road often refers only to police-recorded injury collisions. It cannot necessarily show near misses, a vehicle leaving the road without reported injury, or damage-only events that may still be relevant to engineering concerns. The statistical limitation is not a scandal in itself; many datasets have scope rules. But it matters when a resident says a road feels dangerous and an official response cites very few recorded injury incidents. The two accounts are not necessarily contradictory.

The broader national KSI series remains the right primary source for serious outcomes. The FOIs demonstrate why site-level analysis needs a transparent definition of what enters the collision database. Claims about a road being safe because it has no recorded injury collisions should acknowledge that narrow recording scope and the short periods often involved.

What better scrutiny of the 2030 strategy requires

The next useful evidence release would connect injury severity, road-user type, age, district, road class and principal causation factor for several years in a consistent machine-readable series. It should make clear where data are suppressed for privacy or statistical reliability. For children, reporting should separate fatalities from serious injuries and show the nature of exposure without identifying victims. For rural roads, comparisons should use vehicle kilometres or another defensible measure rather than raw casualty counts.

Another necessary question concerns intervention outcomes. Where 20mph limits, speed enforcement, road engineering or safety education have been introduced, are the recorded KSI trends moving in the intended direction? Can authorities publish before-and-after measures with exposure and uncertainty? Road-safety strategy statistics are designed to make such evaluation possible, but the high-level annual summary does not answer every intervention-level question. This is a reporting agenda, not a claim that PureCar has obtained new unpublished correspondence or interviews.

The fall in deaths deserves recognition, not cynicism. Families spared a fatal collision have avoided an irreparable loss. But families dealing with serious injury also experience permanent consequences that can disappear from a death-focused annual headline. Public authorities should not need to choose which group counts as a success. A credible 2030 strategy has to account for both.

The finding that changes the story

When official records are read together, the central question is not whether 2025 was a good or bad year for road safety. It was better on fatalities and worse on several other core measures. Deaths fell by 13, serious injuries rose by 127, child KSIs rose by 22 and young-people KSIs rose by 20. The strategy recorded serious injuries 42% above baseline and children killed or seriously injured 62% above baseline. Motorcyclists faced a much higher reported KSI rate per distance travelled than car users. These are not contradictory statements; they describe different dimensions of the same problem.

For Northern Ireland's drivers, families and public authorities, the appropriate test is whether the next set of interventions starts narrowing all of those gaps. A lower fatality total is a victory. It is not the final verdict. The more serious investigation begins when the headline is put aside and the people who survived with severe injuries come back into the statistics.

How the investigation was assembled

PureCar examined the PSNI 2025 calendar-year road-collision statistical release, the DfI Road Safety Strategy to 2030 statistical report for 2025 published August 2026, the Road Safety Partnership's 2025 summary, NISRA's analysis of 2019–23 speeding-related serious harm, and PSNI FOI/16713 and another published site-collision disclosure describing damage-only data limitations. All comparisons are labelled by calendar or financial year, age category, causation scope and exposure unit. The figures demonstrate the official recorded pattern; they do not establish the cause of the year-on-year changes or pretend to represent fresh interviews with victims, clinicians or officials.