A decade of NI crash data: why the 2025 rise needs context
PureCar original analysis | 23 October 2026. From 6,225 injury collisions in 2016 to 5,015 in 2025, the long trend differs from the latest annual increase. This report separates recorded evidence from analysis, so readers can understand not only the headline but also the limits of the data.
Start with the full series, not a single year
The official PSNI series puts recorded injury collisions at 6,225 in 2016, 6,081 in 2017, 5,749 in 2018 and 5,676 in 2019. In 2020 the figure dropped to 4,223, followed by 4,704 in 2021, 5,116 in 2022, 5,058 in 2023, 4,753 in 2024 and 5,015 in 2025. The latest annual rise is real, but 2025 remained below each of the four pre-pandemic years shown here.
The pandemic distorts easy comparisons
The year 2020 is a poor standalone baseline for normal traffic patterns because travel behaviour and restrictions were exceptional. A rise from 2020 does not automatically indicate worsening per-mile driving safety. Likewise, an apparent improvement from 2016 is not necessarily a reduction in crash risk per journey. To measure exposure-adjusted risk properly, analysts need compatible traffic-volume, journey or distance data alongside the police collision files.
Casualties moved differently from collisions
The number of casualties was 9,591 in 2016, 9,184 in 2017, 8,720 in 2018, 8,872 in 2019, 6,487 in 2020, 7,192 in 2021, 7,846 in 2022, 7,985 in 2023, 7,494 in 2024 and 7,879 in 2025. The relationship is not fixed: each recorded collision can injure one or several people. That is why a simple crash count is not sufficient when evaluating the burden on families and health services.
How to read the trend responsibly
There are at least three different questions: how many collisions occurred, how many people were hurt, and how dangerous was the road network relative to its use? The PSNI series answers the first two for injury collisions reported to police. It does not by itself supply the denominator required for the third. PureCar is publishing the series as a starting point for local road and junction investigations, not as a stand-alone league table of dangerous routes.
A ten-year view changes the headline
The PSNI historical series gives a useful counterweight to a one-year increase. Recorded injury collisions were 6,225 in 2016, 5,676 in 2019, 4,223 in 2020 and 5,015 in 2025. PureCar calculates that 2025 was approximately 19.4% below 2016 and 11.6% below 2019, despite rising 5.5% on 2024. Those statements can all be true at once.
| Year | Injury collisions | Reported casualties |
|---|---|---|
| 2016 | 6,225 | 9,591 |
| 2019 | 5,676 | 8,872 |
| 2020 | 4,223 | 6,487 |
| 2023 | 5,058 | 7,985 |
| 2024 | 4,753 | 7,494 |
| 2025 | 5,015 | 7,879 |
The drop in 2020 coincides with extraordinary pandemic travel restrictions and changes in movement. Treating that year as a normal baseline distorts the story. Neither the longer fall in collision counts nor the latest rebound measures the rate per mile driven, because those counts have not been adjusted for changes in road use.
About the evidence
This PureCar analysis uses officially published PSNI and Northern Ireland departmental statistics. The police series covers injury road traffic collisions reported to police; it does not capture every non-injury bump, near miss or unreported incident. A collision is an event and a casualty is a person, so the two totals must not be used interchangeably. The figures are historical and do not indicate conditions on any route right now.
What the figures cannot prove
A higher count in a place or group does not, by itself, establish greater risk per vehicle journey, per mile travelled or per person. The appropriate denominator and road-user exposure data would be required for that conclusion. Nor can the published summary alone establish responsibility for individual incidents. Where this article raises questions about infrastructure, enforcement or road-user behaviour, those are reporting questions unless a cited source confirms an answer.
Explore the underlying information
For local road context, see PureCar’s Northern Ireland road safety reports. For present-day road disruption, use PureCar traffic information. The PSNI road collision statistics library contains the official statistical series and supporting files. Figures can be revised by data owners; for definitive tables and definitions readers should check the official publications directly.
