A1 mapped injury collisions fell 43% from the 2014 peak as junction scheme remains pending

The A1 corridor mapped by PureCar recorded 134 police-reported injury collisions in 2025, compared with 236 in 2014, the highest annual count of its 2013–2025 series. The difference of 102 incidents amounts to 43.2% fewer recorded collisions. It is a sizeable change over eleven years, but not a finding that the chance of injury on any particular A1 journey has fallen by 43.2%.

There is a second reason to examine the history carefully: plans to alter junctions on the A1 between Hillsborough and Loughbrickland have been under development for years. The Department for Infrastructure's A1 Junctions Phase 2 project proposes grade-separated junctions, a continuous central reservation barrier and changes to road access. Those details make an engineering debate tangible. Yet the existence of a proposed scheme does not mean the improvement was already in place during the years in which PureCar recorded falling counts.

The central public-interest question is therefore not simply whether the A1's injury-collision tally has dropped. It is what the historical geography reveals, how the remaining serious incidents compare with the overall count, and why a reduction in recorded collisions does not remove the case for examining specific high-consequence junction movements.

What the full thirteen years show

PureCar's A1 mapped corridor recorded 191 injury collisions in 2013, then 236 in 2014. The sequence continued with 190 in 2015, 206 in 2016, 178 in 2017, 200 in 2018 and 194 in 2019. It dropped to 125 in 2020, before registering 140 in 2021, 159 in 2022, 145 in 2023, 151 in 2024 and 134 in 2025. The total across those years is 2,249.

One useful finding is that 2025 was the second-lowest count in the series, above only 2020's 125. That is stronger context than comparing exclusively with the 2014 peak. It places the latest year towards the bottom of the historical range without falsely describing 2025 as an unprecedented low.

The year-on-year fall from 151 in 2024 to 134 in 2025 was 17 incidents, or around 11.3%. But year-to-year movement was not consistently downward: the tally rose from 140 in 2021 to 159 in 2022, fell to 145 in 2023, increased to 151 in 2024 and then declined again. A line drawn from 2014 to 2025 can hide that variation. Looking at every point avoids the temptation to claim an uninterrupted success.

The 2014 count of 236 was 45 higher than the 191 in 2013. The following year, 2015, saw 190, almost a return to the starting figure. That early volatility suggests caution about using 2014 alone as a measure of what was once normal. A longer baseline is a better starting point when making claims about sustained change.

From 2013 through 2018 the mapped corridor recorded 1,201 injury collisions, about 200 per year. From 2019 through 2024 it recorded 914, approximately 152 per year. Those equal six-year periods differed by 287 incidents, or about 23.9%. The later window includes the Covid-19 disruption in 2020, and both periods are unadjusted counts, but the comparison still gives a more stable description than two isolated years.

The moving average helps in the same way. PureCar's rolling three-year annual average peaked at about 210.7 for the period ending in 2016. The equivalent figure ending in 2025 was 143.3. It had edged up to 151.7 at the end of 2024 before declining in the latest calculation. Those figures do not show a perfectly smooth descent; they show a substantial shift in the level of recorded events over time.

The severity record prevents premature celebration

Across the mapped A1 corridor, the police data identify nine fatal-severity collisions, 206 serious-severity collisions and 2,034 slight-severity collisions during the thirteen years. That makes 215 events classified fatal or serious and 2,034 classified slight. Approximately 9.6% of all mapped injury collisions fell into the two most serious incident categories.

The linked casualty count is larger than the incident count: 3,133 people killed or injured. The individual outcomes listed on PureCar's road page are nine people killed, 225 seriously injured and 2,899 slightly injured. These add to the same 3,133 total. There are also 4,086 linked vehicle records, reflecting the vehicles involved across those incidents, not additional separate collisions.

One might be tempted to take the nine fatal-severity incidents and compare them with separate historical reports of A1 road deaths. That would be unsafe without checking how each source defines its geography and period. PureCar's figure covers injury incidents selected within 100 metres of mapped DfI road sections between 2013 and 2025. A reported death toll relating to another set of years or the entire A1 is not necessarily measuring the same thing.

Nor does a falling annual all-injury collision count necessarily mean fatal or serious incidents fell by the same percentage. The public A1 page gives a severity breakdown across the full period, but it does not establish the severity distribution for 2014 versus 2025 within the annual total quoted here. Any claim that fatalities fell 43% on the A1 because the overall count fell by that proportion would be an unsupported leap.

The comparison that can be made safely is more restrained. Recorded nearby injury incidents were fewer in 2025 than at the 2014 high, and the mapped history contains serious human harm. That combination warrants analysis of both frequency and severity, rather than allowing one headline to dominate the other.

Why the proposed junction scheme matters

The Department for Infrastructure's A1 Junctions Phase 2 proposal addresses a defined section between Hillsborough and Loughbrickland. Official scheme descriptions set out four grade-separated junctions intended to enable changes to openings in the central reservation. The named locations are Listullycurran Road, Gowdystown Road, Skeltons Road/Drumneath Road and Waringsford Road. The plan also includes a continuous central median concrete safety barrier.

Other proposed changes include a northbound on-slip at Castlewellan Road near Banbridge, a link road involving the existing Hillsborough Road underpass at Dromore and Milebush Road, and the closure of nine minor road junctions. Remaining local junctions are to operate on a left-in, left-out basis under the design. These are specific design measures intended to change how vehicles turn, cross or join the dual carriageway, rather than merely an announcement to improve driver awareness.

The design explanation is relevant to the way A1 collisions are discussed. A vehicle crossing a central reservation, turning across opposing traffic or emerging through a gap raises distinct engineering questions from rear-end collisions or incidents on adjoining local roads. A grade-separated junction changes the movement of traffic through different levels; closing a median opening removes a particular type of crossing movement. However, the collision totals on PureCar's corridor page do not tell us how many historical injuries were attributable to any one manoeuvre. It would be wrong to reverse-engineer a causal breakdown from the planned design.

The department says the scheme moved to procurement in December 2024 after a July 2024 decision to proceed. Its A1 overview has subsequently noted delays and legal advice arising from the A5 judgment. The department's October 2026 updates on the A5 appeal provide wider legal context, but they do not establish a completed construction programme for the A1 work. The proper status is a proposed improvement at a planning and procurement stage, not a finished intervention that can already be credited with a fall in 2025 collisions.

That sequence matters for public accountability. An A1 article that attributes the entire 2014–2025 decline to a project not yet completed would be a fundamental factual error. Conversely, the existence of fewer recent collisions does not prove that the risk associated with individual crossing movements has disappeared or that the design objectives are obsolete.

A corridor map is useful, but it is not a junction-by-junction safety audit

The PureCar investigation is based on 76 mapped A1 sections. It uses PSNI collision coordinates and DfI road geometry to include recorded injury incidents within 100 metres of the selected road alignment. A collision on an adjacent local road or at a nearby junction may be counted. So may an incident on a connecting road approach that sits within the selection boundary.

This approach helps identify the historical distribution of reported injury collisions around an important transport corridor. It does not confirm that every incident occurred on the A1 mainline. For that reason the most precise expression of the headline is 134 injury collisions near the mapped A1 corridor in 2025, rather than 134 proven collisions on the A1 carriageway itself.

The proximity method also matters when considering the junction proposal. A broad corridor count covering numerous sections cannot show whether historical incidents were concentrated at the four proposed grade-separated junction locations. The map would need to be assessed at the relevant coordinates, with consistent boundaries around each location, before claims about junction-specific trends could be made. No such concentration has been asserted in the headline figures used here.

Similar caution is required if comparing the A1 with the A5 or M1. Each mapped study has a different number of sections and different surroundings. Raw totals would partly reflect how much network is included as well as the incidents recorded. A fair safety comparison needs consistent exposure measures and geography; a numerical league table created from these three summaries would be an editorial shortcut, not a risk analysis.

How much of the fall could reflect different travel patterns?

There is no vehicle-kilometre figure attached to the annual A1 corridor series in the published summary. Without it, one cannot calculate the number of injury collisions per million miles or kilometres driven. The distinction is important over a period spanning changes in commuting, freight movements, public health restrictions and the wider road network.

Injury collisions can fall when traffic falls even if the underlying risk per journey does not change. They can rise when traffic increases even if vehicles are travelling more safely on average. Those possibilities are not excuses to ignore incident counts: every recorded injury event represents actual harm. They are reasons to avoid assigning a change in raw frequency entirely to driving behaviour or engineering.

The 2020 entry illustrates the point. The A1 count fell from 194 in 2019 to 125 in 2020, a reduction of 69. It then increased to 140 in 2021 and 159 in 2022. The unusually low 2020 value should not be presented without acknowledging the changed travel conditions during the pandemic, and the recovery afterwards should not automatically be framed as a failure of road safety policy.

Even outside the pandemic years, a route with major junctions and connections can experience a different balance of local and through traffic over time. A before-and-after study of a specific engineering change would require reliable traffic counts, the date and extent of the change, and matching casualty outcomes. The published 13-year totals alone do not provide that evaluation.

The Department for Infrastructure's 2025 road-safety strategy figures offer broader context: Northern Ireland recorded 56 deaths and 1,066 serious injuries that year. The serious-injury total was 42% above the strategy's baseline figure of 751. A route may show fewer nearby injury collisions than a decade earlier while the jurisdiction still faces serious problems with harmful collisions overall. Neither perspective invalidates the other.

The difference between lower numbers and a safer road

A road's safety can be described in several ways. One is the frequency of recorded injury collisions. Another is how often an incident results in death or serious injury. A third is the risk faced per unit of travel. A fourth, particularly important for a junction project, is whether certain road movements create severe conflicts. The 134-versus-236 headline answers the first question for PureCar's defined mapped corridor; it does not answer all four.

For the public, the most useful conclusion is neither complacency nor alarmism. The longer-run decline is an observable feature of the police record. But the A1 has also recorded deaths and life-changing injuries in the same selected vicinity. A fall in aggregate incidents should not be used to argue that individual locations do not require scrutiny, especially where an engineering scheme has identified problematic crossing and access arrangements.

The converse mistake would be to assume every proposed design element is proven necessary or effective solely because historical harm exists. Assessing the project requires route-specific traffic and collision studies, engineering analysis, the published design and independent scrutiny of the costs and alternatives. The mapped totals are one component of that evidence, not a substitute for it.

The editorial value of the 2013–2025 series is precisely that it resists slogans. It shows a 2014 peak, repeated fluctuations, a distinct fall around 2020 and a 2025 count near the bottom of the displayed range. It also shows 215 fatal- or serious-severity events in thirteen years, a reminder that lower frequency should not be mistaken for the absence of harm. A defensible assessment can acknowledge both simultaneously.

One point in the A1 data is particularly revealing: 2025's 134 nearby collisions were just nine above the unusually low 2020 total of 125. That proximity is notable, but it does not mean the two years experienced comparable levels of traffic or identical circumstances. Read alongside the three-year averages and the equal six-year periods, it supports a limited conclusion—recorded incidents remained near the lower end of the displayed history—without making a causal claim that could not survive scrutiny.

What happens next on the A1

The first task for any updated A1 assessment is to separate the relevant improvement scheme's specific junctions from the much larger mapped road corridor. That would allow a genuine baseline for the road movements the scheme is intended to change. A future comparison could then distinguish incidents on the through route, local approaches and turning movements, rather than relying solely on an aggregate that includes many types of road environment.

A second task is to track the department's stated programme against decisions actually made. Procurement activity, statutory orders, preparatory work and physical construction are not interchangeable milestones. Until a junction intervention is delivered, a reduction in historical collisions cannot properly be credited to it. Once it is delivered, any assessment of effectiveness should include serious casualty outcomes and changes in vehicle movements.

For now the verified record is clear. The mapped A1 corridor counted 134 police-reported injury collisions in 2025, the lowest annual tally except for 2020, and 43% below the 2014 peak. It recorded 2,249 such incidents over thirteen years, with nine people killed and 225 seriously injured in linked casualties. A defined A1 junction programme remains a separate question of proposed future road design and public investment.

Publishing those facts together is more useful than declaring the road safe or dangerous from a single total. Readers deserve to know both what has improved in the recorded trend and what has not yet been demonstrated about the reasons for it.

Evidence used

The corridor figures and annual series are published in PureCar's A1 Collision History and Road Safety investigation, which uses a 100-metre match to DfI road geometry. The project's scope and status are documented in the Department for Infrastructure's A1 Junctions Phase 2 overview. The wider annual casualty context is in the PSNI 2025 Key Statistics Report. The study concerns police-reported injury collisions and excludes unreported and damage-only events.