Research, Pass rate vs population density

UK driving test pass rate vs urban density, the gradient generalises

Across the 273 UK car test centres with at least 500 tests in 2024-25, the urban-core tier (postcode areas anchored in inner London, Birmingham, Manchester, Liverpool, Glasgow, Leeds, Sheffield, Newcastle, Bristol, Cardiff, Edinburgh and other major city centres) passes at 47.8%. The rural tier (the Scottish Highlands and Islands, mid-Wales, far-west Cornwall) passes at 51.5%. That is a 3.7-percentage-point gap and the gradient is monotone across all four density tiers. Pearson correlation between the density ordinal and per-centre pass rate is -0.201, a moderate negative relationship that holds inside England, Scotland and Wales separately. Headline figures from the DVSA DRT122A release for 2024-25, with the density classification anchored in ONS Census 2021 built-up-area population density.

By VikasPublished15 May 2026Methodology Sources

UK volume-weighted

48.6%

273 car centres, 2024-25

Urban-core tier

47.8%

75 centres, 598,531 tests

Rural tier

51.5%

20 centres, 52,833 tests

Urban-to-rural gap

3.7pp

Rural minus urban-core

Pearson r (density ordinal)

-0.201

Moderate negative correlation

Density tiers

Urban core / suburban / market town / rural

Section 1, The gradient is monotone across four density tiers

The headline number on this page is a 3.7-percentage-point gap in pass rate between the rural and urban-core tiers in the latest DVSA release. That gap, taken alone, could be anything from a clean signal to a noisy artefact of how rural and urban centres happen to sit on the map. The chart below shows it is a signal. Each of the four tiers passes lower than the rural tier and the gradient never reverses.

The rural tier carries only 20 centres and 52,833 tests in the latest year, so its headline rate has a wider confidence band than the others. But the next tier up (market town: 88 centres holding 465,615 tests at 50.1%) still sits above the UK average, and the major-suburban tier (90 centres at 48.1%) sits below it. The urban-core tier at the top of the chart carries 75 centres and 598,531 tests at 47.8%. The middle of the distribution is dense, the tails are populated, and the gradient is consistent.

Section 2, How urban density was measured, postcode areas as the density unit

Density per centre is not directly published in the DVSA data, so we use a proxy. Each centre's full street address resolves to a UK postcode, and the postcode's alphabetic area letters (the "outward" prefix; SW, B, PA and so on) bucket the centre into one of the 124 UK postcode areas. Each postcode area has a well-documented population density at the centre of its dominant built-up zone, anchored in ONS Census 2021 tables. We collapse the 124 areas into four density tiers:

Urban core (75 centres, 5,000+ /km² (typical built-up zone)): postcode areas whose centre of population sits in a built-up zone with very high density. Includes London's compass postcodes (E, EC, N, NW, SE, SW, W, WC) plus the inner cores of Birmingham (B), Manchester (M), Liverpool (L), Glasgow (G), Leeds (LS), Sheffield (S), Newcastle (NE), Bristol (BS), Cardiff (CF), Edinburgh (EH), Bradford (BD), Leicester (LE), Nottingham (NG), Coventry (CV) and Portsmouth (PO).
Major suburban (90 centres, 1,000 to 5,000 /km²): outer-London commuter belt (BR, CR, DA, EN, HA, IG, KT, RM, SM, TW, UB), the Greater Manchester ring (BL, OL, PR, SK, WA, WN), the Black Country (DY, WV, WS, ST), Thames Valley (RG, SL), Home Counties (MK, LU, AL, WD, SG, GU, RH, BN, ME, CT, TN, SO, BH) and several other dense-suburban areas.
Market town / mixed (88 centres, 200 to 1,000 /km²): the smaller English cities and market towns. Includes BA, CB, CO, IP, NR, PE, LN, GL, OX, SN, SP, DT, EX, PL, TQ, TA, CW, LA, FY, CA, DH, SR, TS, DL, HG, YO, HU, plus most Scottish lowland centres (AB, DD, FK, KY, KA, ML, PA), Northern Ireland (BT), Newport-Gwent (NP) and mid-Wales (LD).
Rural (20 centres, <200 /km²): the genuinely low-density areas. Scottish Highlands and Islands (HS, IV, KW, PH), Dumfries and Galloway (DG), the Borders (TD), mid-Wales coast (SA), north-west Wales (LL) and the far Cornish coast (TR).

The 124-to-4 collapse is intentionally coarse: we want a gradient that survives reasonable alternative groupings rather than one that is sensitive to where any single boundary falls. Alternative groupings (six tiers, ONS quintile cuts, separate "London core" + "other major city" tiers) all produce the same monotone shape with slightly different headline numbers.

Section 3, Why the density gap is smaller than the volume gap

The volume-vs-pass-rate research page, shipped one day before this one, documents an 11.5-percentage-point gap between the lowest and highest volume tiers in the same dataset. This page documents a 3.7-percentage-point gap between the lowest and highest density tiers. Both gradients are monotone, both correlations are negative, and both stories agree about the underlying mechanism. So why is one gap roughly three times the size of the other?

Two reasons. The first is that postcode area is a coarse density proxy. An "urban core" centre in the SW postcode area can sit in dense Wandsworth or in much less dense Wimbledon Common, and we're averaging across both. The candidate doesn't drive across the whole postcode area; the candidate drives on a 35-minute test route inside it, and the route chosen by the centre might sample the dense end or the quieter end of the area. Averaging across the postcode area washes out within-area variation that the centre's actual test routes preserve.

The second reason is that centre annual volume is a sharper signal because it tracks the catchment a centre actually serves. A centre placed in a sparse part of an "urban core" postcode area will be lower-volume than a centre placed in a dense part of the same area, because the catchment is sparser; the volume number captures that directly, the density number does not. Volume is acting as a per-centre density proxy and density-per-postcode-area is acting as a per-area average. The per-centre proxy is sharper.

The two analyses agree on the underlying claim (denser environments produce lower pass rates) and disagree on the magnitude of the headline gap (volume cuts produce sharper tiers than density cuts). The honest reading is that volume is the better operational lever and density is the better structural framing. A learner choosing between two real centres should look at volume; a journalist or researcher framing why London passes lower than rural Scotland should look at density.

Section 4, The gradient holds inside every UK region

A natural concern with a four-tier density classification is that the gradient could be a sleight-of-hand version of the London-vs-UK pattern. The density tiers happen to put most London centres into one bucket (urban core) and most rural Scottish centres into another (rural), so the headline gap might be the London-vs-rural gap rebadged. The chart below tests that. It shows pass rate by UK region for the regions with at least 3 centres in the analysed set, with the sublabel showing each region's share of tests in urban (urban-core plus major-suburban) postcode areas. The regions with high urban-share pass low; the regions with low urban- share pass high. The pattern is not London-specific.

Within Scotland, the urban-core Glasgow and Edinburgh centres pass below the rural Highland and Islands centres by a similar gap to the headline four-tier gradient. Within Wales, the urban-core Cardiff centres pass below the rural Mid-Wales and North-West Wales centres. Within England, the same pattern holds with the urban-core inner-London and Birmingham centres at the bottom and the rural East Yorkshire, Cumbria and Cornwall centres at the top. The density gradient is not a London oddity; it is a UK-wide structural pattern.

Section 5, A typical urban-core centre vs a typical rural centre

Put one representative urban-core centre alongside one representative rural centre and the structural difference is direct.

Birmingham (Garretts Green)

Region: West Midlands
Postcode area: B
Density tier: Urban core
Annual tests: 21,871
Pass rate: 42.0% (2024-25)
vs UK average: -6.6pp
Route mix: dense urban built-up zone, multi-lane roundabouts, bus lanes, pedestrian crossings, mandatory dual carriageway

Bala

Region: Wales
Postcode area: LL
Density tier: Rural
Annual tests: 574
Pass rate: 61.1% (2024-25)
vs UK average: +12.5pp
Route mix: rural A-roads, single-carriageway, low traffic density, fewer mandatory dual-carriageway segments

Same DVSA syllabus, same examiner training pipeline, same DT1 marking sheet, same legal test format. The urban-core centre sits in a postcode area roughly 11x denser by population than the rural centre and lands 19.1 percentage points below it on pass rate. The two tests measure the same skills and produce very different outcomes because the environments they measure those skills in are structurally different.

Section 6, What density is a proxy for, and what it isn't

Population density per postcode area does not, by itself, mark down a candidate. The mechanism is downstream of density: dense postcode areas have dense road networks, and dense road networks present more faultable situations per test minute. The DVSA examiner using the same DT1 marking manual at a dense centre finds more situations to assess than the same examiner at a sparse centre, because more situations occur. Density is a proxy for: more junctions per mile, more multi-lane roundabouts, more bus lanes, more pedestrian crossings, more cyclists, more buses, more unexpected lane changes from neighbouring cars.

Density is not a proxy for examiner severity. DVSA's examiner training is centralised at Cardington, examiners rotate across centres, and within-examiner consistency audits do not show urban examiners marking systematically harder than rural ones. Density is also only weakly a proxy for candidate readiness. Inner-city learners may have less consistent private practice access (no driveway, no parental second car), but the size of that effect is small compared with the route-environment driver, and it is partially offset by the denser instructor pool urban areas typically have.

The honest framing is: density is a coarse but defensible summary statistic for "how complex is the road environment the candidate will be tested on", and the pass-rate gradient this page documents reflects that complexity, mediated by the DT1 marking sheet. The volume-vs-pass-rate page covers a similar story with a sharper proxy. The London-vs-UK page covers the headline implication for an inner-London learner in particular.

Section 7, What this means for a candidate choosing a centre

The density gradient is real and the temptation to "shop" centres is real. Three points worth keeping clear before chasing the gradient.

The gradient is a population statistic, not a personal one. A learner who can drive an inner- London multi-lane roundabout without faulting will pass at the urban-core centre too. The gradient describes the average candidate's outcome at the average centre in the tier, not the individual's. Use the centre rate as a backdrop, not a prediction.

Density correlates with wait-time, which compounds. Urban-core centres tend to have the longest waits in addition to the lowest pass rates. Our wait-time by region research page quantifies the wait gap; this page quantifies the density and pass-rate gap. Reading both together gives a candidate in dense urban Britain the full decision space: stay local and face longer wait plus lower pass rate, or travel to a less dense centre and trade a familiar environment for a shorter wait and higher pass rate.

Route familiarity is the binding constraint on travelling. A candidate who trains in inner London and books a test at a rural Scottish centre is taking a meaningfully different test. The pass-rate uplift is real and the test is different at the same time. The standard mitigation is one or two instructor-led lessons in the destination centre's area before the test date, which the on-site guide on whether to travel for an easier test covers in detail.

Section 8, Methodology and limitations

Data source. All centre-level pass-rate figures come from the DVSA DRT122A release ("Car driving tests conducted, passed and pass rates by test centre and gender"), licensed under the Open Government Licence v3.0. The latest year covered in this analysis is 2024-25, the most recent complete financial year DVSA has published at the time of writing.

Density classification. Each centre's postcode area letters (the alphabetic prefix that opens a UK postcode) are extracted from the centre's full OSM-style display address. The 124 UK postcode areas are then mapped into four density tiers based on the ONS Census 2021 built-up area population density tables and the ONS Rural- Urban Classification 2011 for areas not contained in a major built-up zone. Tier assignments use the centre of population of each postcode area's dominant built-up zone, not the area's geographic centroid; this matters for areas that span urban + rural (NE, NG, LE) where the centre of population is urban but the geographic centroid would be much more rural. The lookup table is hand-curated and the full tier assignment is exported alongside the helper for reproducibility.

Centre filter. Same filter as the rest of the site: motorcycle-only ("(R)") and heavy ("LGV", "MPTC") centres excluded, the project's KNOWN_CLOSED set of permanently closed centres dropped, and any centre id starting with "z-" dropped. Within that set, we restrict to car centres with at least 500 tests in 2024-25, the same rankable floor used by the London-vs-UK research page and the easiest/hardest rankings. The filter yields the 273 centres counted in the hero stat-grid. The postcode-area lookup achieved 100% coverage of the analysed set, with no centres dropped for unresolved postcode.

Tier-rate methodology. Tier rates are volume-weighted within tier (sum of passRate × tests divided by sum of tests); medians are the unweighted median of per-centre pass rates within the tier. Per-region rates in section 4 use the same volume-weighting formula. Pearson correlation is computed on the density-tier ordinal (1 for rural, 2 for market town, 3 for major suburban, 4 for urban core) vs per-centre pass rate, across all 273 analysed centres. We report three decimal places.

What this analysis cannot do. The page does not establish causation. Density per postcode area is a proxy for the route environment, not a measure of it. A more definitive analysis would use route-level data on the actual junctions, multi-lane roundabouts and dual- carriageway minutes per centre's test routes; DVSA does not publish that. The density proxy is the best we can do with public data, and the page is honest that the volume-vs-pass-rate analysis (with a different proxy) tells essentially the same story with a sharper headline.

Cite this page: passrates.uk research/pass-rate-vs-population-density v1.0 (2026). Data: DVSA DRT122A "Driving test and theory test data, cars" for 2024-25, OGL v3.0. Density: ONS Census 2021 built-up-area population density tables, ONS Rural-Urban Classification 2011. Analysis: 273 UK car centres with at least 500 tests in 2024-25, classified into 4 density tiers via postcode area lookup. Pearson r (density ordinal vs pass rate) = -0.201.

Frequently asked questions

Does urban population density predict UK driving test pass rates?

Yes, the relationship is monotone but moderate. Across the 273 car test centres analysed for 2024-25, the urban-core tier (postcode areas covering inner London plus the inner cores of Birmingham, Manchester, Liverpool, Glasgow, Leeds, Edinburgh, Cardiff and the other major English cities) passes at a volume-weighted 47.8%. The rural tier (postcode areas covering the Scottish Highlands and Islands, mid-Wales, north-west Wales and far west Cornwall) passes at 51.5%. The gap is 3.7 percentage points, and every intermediate tier sits in between. Pearson correlation between the density ordinal and per-centre pass rate is -0.201, a moderate negative relationship.

What density classification did you use, and where did the tier boundaries come from?

We used UK postcode areas (the alphabetic prefix that opens a postcode, e.g. SW, B, PA) as the density unit, then collapsed them into four tiers anchored in ONS Census 2021 built-up-area population density tables and the ONS Rural-Urban Classification for areas not covered by a major built-up zone. Urban core captures postcode areas whose centre of population sits in a built-up zone with density above 5,000 per square kilometre. Major suburban captures the 1,000 to 5,000 band (outer London, the Greater Manchester ring, the Black Country, the Thames Valley and the Home Counties). Market town covers the 200 to 1,000 band (most English county towns, peripheries of larger conurbations). Rural covers the under 200 per square kilometre tier (Scottish Highlands, mid-Wales, Cornish far-west). The boundaries are intentionally coarse so the headline gradient survives reasonable alternative groupings.

Why is the urban-rural gap on this page smaller than the volume-based gap on the related volume-vs-pass-rate page?

Two reasons. Postcode area is a coarse density proxy: an "urban core" centre in the SW postcode area can sit in dense Wandsworth or in much-less-dense Wimbledon Common, and we're averaging across both. Centre annual volume is a sharper signal because it tracks the catchment a centre actually serves: a Wimbledon Common centre will be lower-volume than a Wandsworth one because the catchment is sparser. The volume-vs-pass-rate page documents an 11.5-point gap between the lowest and highest volume tiers; this page documents a 3.7-point gap between the lowest and highest density tiers. Both gaps are real, both gradients are monotone, but volume is the cleaner proxy for the "what is the candidate driving through" mechanism. Density is the cleaner proxy for the "what kind of population catchment is the centre embedded in" mechanism. The two mechanisms compound.

How is "urban core" defined and why does it include centres outside London?

Urban core means a postcode area whose dominant built-up zone has population density above 5,000 per square kilometre, measured at the centre of population. That covers London's compass postcodes (E, EC, N, NW, SE, SW, W, WC) but also B (Birmingham core), M (Manchester core), L (Liverpool core), G (Glasgow core), LS (Leeds), S (Sheffield), NE (Newcastle), BS (Bristol), CF (Cardiff), EH (Edinburgh) and several other major-city centres. The story this page tells is that the London-vs-UK pass-rate gap is not a London oddity; it's a density gradient that runs across every major UK city core. The London-vs-UK research page covers the London-specific case with a sharper filter; this page generalises it.

How strong is the correlation between density tier and pass rate?

Pearson correlation between the density-tier ordinal (1 for rural, 2 for market town, 3 for major suburban, 4 for urban core) and per-centre pass rate is -0.201. That is a moderate negative relationship: density explains roughly 4% of the variance in centre-level pass rates. The rest is absorbed by factors that density is a noisy proxy for: route mix at the specific centre, traffic density on the test routes (not the whole postcode area), examiner supply, candidate selection. The coefficient is large enough to be operationally relevant for a learner choosing a centre and small enough that density alone never determines an outcome.

Should I travel from a dense urban centre to a less dense rural one for a higher pass rate?

It is a real lever and a misunderstood one. The pass-rate gap between an urban-core centre and a rural one is 3.7 percentage points on average, the wait-time gap can be a few weeks one way or several months the other, and a candidate who is genuinely ready can sometimes book a faster test at a higher-passing rural centre by travelling. The catch is route familiarity. A learner trained on dense inner-city routes who turns up at a rural centre is taking a different test in practice: narrower lanes, fewer traffic lights, unfamiliar speed-limit road layouts, livestock and farm vehicles. The pass-rate gap is not a friendlier examiner; it is a structurally different driving environment. The on-site guide on whether to travel for an easier test covers the calculus in detail, and the standard mitigation is one or two instructor-led lessons in the destination centre's area before the test date.

Does density correlation hold within Scotland and Wales as well as England?

Yes, the per-region cross-cut at the bottom of this page confirms it. Within Scotland, the urban-core Glasgow / Edinburgh centres pass several points below the rural Highland and Islands centres. Within Wales, the urban-core Cardiff centres pass below the Mid-Wales and North-West Wales rural centres. The pattern is not country-specific; it is a property of how dense, route-complex urban catchments produce more faultable driving situations per test minute than sparse rural ones do. England, Scotland and Wales each show the same internal gradient.

Are there exceptions, dense urban centres that pass at or above the UK average?

A small number, and the pattern of those exceptions is informative. Several urban-core centres in commuter-belt or fringe city areas pass at or above the UK average of 48.6% (Manchester (West Didsbury) at 50.9%, Cardiff (Llanishen) at 51.3% are examples). The common thread is that these centres sit in postcode areas whose dominant built-up zone is dense but whose test routes actually sample a quieter mix of suburban arterial roads and modest roundabouts, not the inner-core complexity. The exceptions reinforce rather than contradict the main pattern: density at the postcode-area level is correlated with pass rate, but the binding constraint is what the test route asks the candidate to drive through, and that varies within a postcode area.