Northern Ireland vehicle crime fell 8.7% — but the final months tell a different story

Northern Ireland's latest mapped vehicle-crime figures contain two stories that appear to point in opposite directions. Across the 12 months from September 2025 to August 2026, the total was lower than in the preceding year. Yet, at the end of that period, the monthly count was heading sharply upwards. Neither observation cancels out the other. Together, they show why a single percentage cannot adequately describe the experience of vehicle crime across the region.

PureCar's Vehicle Crime Northern Ireland database contains 2,152 police-mapped vehicle-crime records for September 2025–August 2026. The comparable September 2024–August 2025 window contained 2,357. That is a decrease of 205 records, or 8.7%, in this particular published and mapped series. The database covers 36 months, from September 2023 to August 2026, across Northern Ireland's 11 council areas. Its 36-month historical view is more useful than a headline that isolates a single good or bad month.

The reduction is significant enough to merit attention. But the same series recorded 228 vehicle-crime entries in August 2026, its largest monthly count in the latest 12 months. Just two months earlier, the June total was 145. For drivers reading a report about falling crime while watching the latest monthly numbers rise, that distinction matters. A better annual comparison does not mean every month was safer, and an unusually busy month does not prove that the annual trend has reversed.

The year in 12 figures

A month-by-month examination of the latest year reveals how unevenly the 2,152 entries were distributed:

The average for the period was approximately 179 entries a month. That average is useful as a reference point, but it describes almost none of the months well. November, December, January and June were all considerably below it. March, April, July and August were clearly above. A resident could therefore experience a period of repeated incidents locally at the same time as the Northern Ireland-wide yearly total declined.

The most revealing comparison is between the first and second halves of the latest 12-month period. September to February accounted for 974 entries. March to August accounted for 1,178, a difference of 204 records, or approximately 21%. That does not establish a seasonal trend: identifying seasonality would require comparing the same months across several years, accounting for reporting changes and checking the longer monthly series. It does, however, demonstrate that the lower annual total was not distributed evenly throughout the year.

A late-summer rise is not the same as a crime wave

July's 206 records were up from 145 in June. August increased again to 228. Those are material month-to-month movements, but describing them as a 'crime wave' would go beyond what these figures can establish. Monthly movements can reflect a variety of influences, including fluctuations in offending, reporting, police recording, and the timing of data publication or revision. A map count cannot on its own identify the cause.

It is also important to distinguish police-recorded, mapped vehicle-crime entries from the number of individual cars stolen. The vehicle-crime category covers theft of or from a vehicle and interference with a vehicle. A record might relate to property taken from a parked car or an attempted offence rather than a complete theft of the car. The number 2,152 is not a stolen-car count, nor does it establish the number of unique victims, total financial losses, insurance claims or vehicles never recovered.

This is not a technical quibble. The practical questions differ. A driver whose entire vehicle is taken may face a loss of transport, claims handling and the risk of vehicle identity fraud. Someone targeted by a theft from a van may face a very different disruption, especially where equipment needed for work has gone. Vehicle interference may involve damage without a completed theft. A single category can describe those offences collectively for mapping, but it cannot make their consequences identical.

What the longer history adds

PureCar's archive reaches back to September 2023 and is presented through the 2026 vehicle-crime report. It has been reconciled across all 11 council areas and 36 monthly reporting periods. That matters because comparisons must use the same boundary, the same category and matching periods. Taking the August total from one system and a separate official car-theft total from another is not a like-for-like comparison.

There is an equally important geographical limitation. The underlying Police.uk street-level data deliberately obscures incident addresses. Locations on a map are approximate, and some incidents without a suitable mapped location are not displayed. Consequently, the latest mapped total should not be presented as a complete census of everything that might be reported to police under every vehicle-related offence code. It is a consistent measure of what appears in this particular mapped public dataset.

The 8.7% fall is therefore a defensible finding about PureCar's mapped series, not proof that the likelihood of any individual vehicle being targeted has fallen by 8.7%. To estimate personal risk, researchers would need additional measures: the number and types of vehicles in use, where and how long they are parked, local exposure, reporting rates and offence-specific denominators. Those are not supplied by the map total.

What a Northern Ireland driver should take from the figures

The strongest use of these statistics is neither reassurance nor alarm. It is to show where questions are needed. Why did the monthly number strengthen in March and April? What explains the June dip followed by the July–August increase? Did that pattern occur in the same council areas, or did different places contribute at different times? Are thefts from vehicles moving differently from complete vehicle thefts? The published aggregate alone does not answer those questions.

Drivers should not change basic prevention habits based on a single falling percentage. The PSNI advises owners to secure keys and fobs, check that electronic locking has worked and keep valuables out of view. Vehicles fitted with keyless entry may benefit from signal-blocking storage and a visible physical security device. Those precautions are relevant whether the latest monthly headline is better or worse than the one before it.

For anyone who becomes a victim, prompt reporting is essential. PSNI advises people to contact police as soon as they become aware that a vehicle has been stolen, damaged or entered, and to avoid disturbing possible forensic evidence. Clear vehicle details, approximate times and preserved camera footage can help investigators; the published map cannot replace a formal police report.

The arithmetic behind the annual improvement

The easiest mistake in evaluating the headline fall is to put the latest monthly reading alongside the previous year's total and treat the two as comparable. PureCar's comparison does not do that. It totals 12 consecutive monthly entries in one window, September 2024–August 2025, and compares that result with the next 12 consecutive monthly entries, September 2025–August 2026. Both windows have the same number of calendar months, with no overlap.

Subtracting 2,152 from 2,357 gives a difference of 205. Dividing 205 by the earlier 2,357 produces approximately 0.087, hence the reported 8.7% reduction. The percentage describes the proportional change in the recorded and mappable dataset; it does not imply that precisely 205 fewer people became victims. A person can experience multiple offences, while a recorded offence is a unit governed by counting rules rather than a census of people.

The monthly table provides another check. Adding the 2025 months from September through December gives 657 records. Adding January through August 2026 gives 1,495. Together they total 2,152. This is an unusually useful verification step because the reader can check the sum without taking the headline on trust. Where a report supplies only a percentage without a denominator, a large-looking change can conceal a small count. Here the denominator is explicit.

The downward comparison would also look different if the selected dates changed. A rolling year ending in August is not a fixed financial year and not a calendar year. That is not a defect: rolling periods are often useful for tracking newly published monthly data. The condition is that the comparison must keep those windows aligned. The reason PureCar specifies both start and end months is to prevent an analyst from accidentally comparing 11 months with 12 or shifting one period to include an unusual month.

What would count as evidence of a sustained reversal?

A rise in July and August raises a question, not a settled finding. To establish that the broader decline has ended, analysts would look at subsequent monthly releases and then compare each new rolling-year total with the 12 months before it. They would ask whether a climb persists after temporary changes and data revisions, whether multiple districts show it, and whether particular offence types are responsible.

There is also a difference between a month reaching a high level and a persistent higher rate. August recorded 228 mapped entries against an annual monthly average of roughly 179. It exceeded that average by about 49 entries. March, April and July had already been above the average, so August did not emerge from a flat or steadily falling sequence. Conversely, the lowest month, June, demonstrates that a single dip can be misleading too. The whole run of observations is necessary to understand the scale of variation.

The earlier 12-month count of 2,357 gives a reference point for the annual reduction, but the monthly figures of that earlier window would be needed for a true August-on-August comparison. Without putting those corresponding months side by side, it would be wrong to write that August 2026 was higher than August 2025 or to claim a seasonal record across the full archive. The current article makes the narrower and verified statement: August was the largest month within this particular latest 12-month period.

A stronger understanding would also depend on changes in how incidents are classified. Vehicle crime is not a single type of offender behaviour. A fall in attempted thefts alongside a rise in vehicles actually taken would carry different consequences from the reverse. The public map's broad category is valuable for the overall geography, but it cannot answer the more specific question without offence-level records. That limitation is why PureCar also identifies separate PSNI offence tables as a distinct source.

How a local reader can test a headline

For a resident looking at a district page, the first useful step is to confirm which month or year the interface is showing. A map set to August 2026 will tell a very different story from a map set to a whole 12-month window. A count shown at one anonymised location should then be treated as a mapped grouping, not as a list of incidents at a door. A large council total may cover many towns and thousands of parking places.

A second step is to examine direction over several months. If a council's count rises one month, that might be notable but could also be random fluctuation against a low baseline. A sequence across several months is more informative than a single point, although even a repeated pattern does not reveal motive. A third step is to check the source: this map contains Police.uk/PSNI vehicle crime, not necessarily every vehicle stolen or every insurance claim in Northern Ireland.

The public is entitled to ask local policing partnerships for finer, comparable information where privacy permits. Are thefts of vehicles and thefts from vehicles changing differently? Is a particular type of parking site involved? Are there recoveries or preventative interventions that can be evaluated? Such questions are more useful than labelling an entire town a hotspot because its share of one year's total moved.

A map is a measure of reports, not the full burden of harm

Recorded offences cannot describe the complete effect of crime on a family or business. Losing access to a vehicle can affect commuting, childcare, caring responsibilities and income. Losing tools can stop a tradesperson working, even when the vehicle remains on the road. Those consequences deserve reporting but would require victim evidence or reliable claims data. They cannot legitimately be calculated from the number of coloured map points.

Equally, a fall in the mapped series does not prove that criminals have become less active. It reports what has entered the published, mappable police system. Changes in willingness to report incidents or in the ability to give an exact enough location might influence mapped totals. A robust investigative account names those possible limitations, then states clearly that the data alone cannot determine which, if any, caused the change.

The improvement should therefore be described neither as a triumph nor as an illusion. It is a measured decline in a defined series. Its public value lies in giving residents an intelligible baseline while pointing to the questions not yet answered by the available evidence.

One further check belongs in future reporting: any revision to a previous month should be reflected in both the latest and comparison totals before publication. Police datasets can be revised as information is validated. A figure copied into a draft weeks earlier should not be assumed to remain the latest version. That is especially important for local pages, where a change of only a handful of records may alter the apparent direction of a month. PureCar’s archive is useful because it provides a repeatable period definition; maintaining that value requires keeping the displayed dates, calculations and limitations consistent whenever the figures are refreshed.

The monthly spread also highlights the value of preserving the underlying table when charts are updated. The published rise from June to August can be checked against the monthly figures of 145, 206 and 228. The increase across those two monthly steps is 83 records, but that is not 83 confirmed additional stolen cars. The distinction between a change in mapped offence entries and a change in completed vehicle theft is essential to every subsequent claim about the series. Clear tables protect against that kind of accidental exaggeration.

The finding that should shape future coverage

There is genuine positive news in the year-on-year reduction: 205 fewer mapped records than in the preceding comparable 12 months. But a responsible newspaper should not turn that improvement into an unqualified claim that Northern Ireland's vehicle-crime problem is disappearing. The last three months alone—June, July and August—show how volatile the series remains.

The next test is whether future monthly releases confirm the late-summer movement or show it settling back. Until then, the fairest conclusion is specific: PureCar's police-mapped vehicle-crime count fell by 8.7% over the latest comparable year, while August 2026 was the highest month within that year. Both facts deserve space in the headline story, because readers need the entire pattern rather than the half that looks most dramatic.

Data note: PureCar analysis of its reconciled Police.uk/PSNI vehicle-crime map archive, 36 months to August 2026. The 12-month comparison uses September–August in both years. Police.uk's broad vehicle-crime map category and anonymised location rules apply; figures should not be equated with exact-address incidents or car theft alone. Detailed source and boundary explanations are in PureCar's methodology.