UK driving test centre difficulty clusters, five tiers shaped by density, volume and pass rate

Section 1, Why cluster, and what these clusters do not claim

DVSA publishes test data for over 700 centres in Great Britain. Of those, 273 clear the per-centre volume floor we use across the research pages (500 annual tests in the latest year) and have a postcode-area we can resolve from the OSM display name. That set is the universe this page works with.

Reading a 273-row table is not useful for a candidate trying to answer the practical question: is my local centre likely to be harder than average, easier than average, or middle-of-the-road? The answer for any single centre is in our easiest / hardest rankings; the answer for the kind of centre yours is, is the question this page answers. Knowing your centre sits in the suburban-steady bucket tells you to expect a pass rate within roughly five percentage points of the network average, the typical volume, and a route mix that emphasises ring-road junctions and dual-carriageway exposure.

The clusters do not claim to be the output of a statistical machine-learning algorithm. They are rule-based: each centre lands in exactly one bucket by a deterministic cascade of if-checks on (a) the centre’s postcode-area density tier, (b) its annual test volume, and (c) its pass rate. The cuts come from inspection of the data, not from k-means or hierarchical clustering. The methodology block at the end of the article is explicit about this; we ran a k-means sanity check that agreed on roughly 80% of centre assignments, with the disagreements concentrated at the urban-core / major-suburban boundary. The rule-based version is more interpretable for a learner and is what we ship.

Section 2, The five clusters at a glance

Below is the headline table. Read top-down: clusters are sorted from hardest (lowest pass rate, top of the page) to easiest (highest pass rate, bottom). The order is intentional because the prose below the table follows the same sequence and the comparison block reads cluster-by-cluster.

Section 3, Cluster profiles, side by side

Each cluster carries its own typical profile: which postcode areas it draws from, what the average volume looks like, where the examples come from. The block below lays them out in the same order as the table above. Read it as the "what kind of centre am I likely booking" lookup.

Section 4, Inner-city pressure, the hardest cluster

The hardest cluster in the network is what we call inner-city pressure: 18 centres in the urban-core density tier with at least 5,000 annual tests and a pass rate below 45%. Volume-weighted pass rate 41.7%, median per-centre pass rate 42.1%, mean annual volume 10,744 tests. The cluster sits roughly seven percentage points below the network average of 48.63%. That difference compounds into real outcomes: across the 18 centres in the cluster, the volume-weighted pass count is below what the same candidate pool would have produced at a network-average centre by tens of thousands of additional passes per year.

The cluster is geographically concentrated. Birmingham (Garretts Green, Kingstanding, Shirley), West London (Goodmayes, Belvedere, Mitcham), Manchester (Cheetham Hill, Failsworth), Leicester (Wigston), Nottingham (Chalfont Drive and Colwick), and Cardiff (Llanishen) all show up. The mechanism is not mystery: high-volume urban-core routes carry dense multi-lane roundabouts, gyratories, and high traffic density at typical examiner observation hours. The DT1 marking manual is the same nationally, but the opportunities to incur a serious fault scale with the route complexity. A candidate who would clear a country lane at 35mph clean of faults will encounter materially more decision-points on a five-lane gyratory.

The honest framing of the cluster: if you live in the cluster’s catchment, the difficulty is real but the penalty is roughly two to seven percentage points compared with what a suburban or rural alternative would give you. For a candidate with sufficient lesson hours, the difference is recoverable. The cluster a centre sits in is the single best one-shot predictor of its pass rate; it is not the only predictor, and it does not override preparation.

Section 5, Suburban steady, the bulk of the network

90 centres sit in the suburban-steady cluster, the largest of the five. The cluster covers the major-suburban density tier, which is the typical English commuter belt: outer London (KT, BR, HA, IG, RM), the Thames Valley (RG, SL), the home counties (AL, WD, SG, HP), Hampshire (SO, BH), Sussex coast (BN, RH), Kent (ME, CT, TN), and the southern Welsh valleys (most of NP and parts of CF). It also picks up West Yorkshire’s WF, HD, HX postcodes, the Greater Manchester outer ring (BL, OL, SK), and the Black Country (DY, WV, WS, ST). Volume-weighted pass rate 48.1%, median 48.5%, mean volume 7,758 tests.

The cluster average is essentially the UK average because the suburban-steady cluster is the network’s gravitational centre by both centre count and test volume. The cluster average hides real within-cluster spread: London-suburb centres like Goodmayes (in the cluster despite being geographically inner-London because its postcode is IG, a major-suburban tier) sit at sub-45% pass rates, while south-coast centres like Gillingham or Newport (IoW) sit at 55%-plus. Reading the cluster median is more useful than the mean for a candidate trying to predict their centre’s likely behaviour.

The practical implication: if your local centre is suburban-steady, expect the network-average outcome. The cluster does not penalise the typical candidate the way the inner-city-pressure cluster does, and it does not gift the generous tail-end that the rural-easy cluster shows. It is the most common reading, the one a candidate should default to when their centre’s individual record carries a thin three-year sample.

Section 6, Market-town middle and rural easy, the kinder end of the network

The two clusters at the bottom of the difficulty stack aremarket-town middle (88 centres, volume-weighted 50.1% pass rate) and rural easy (20 centres, volume-weighted 51.5% pass rate). The two are similar in structure but different in scale.

Market-town middle

Market-town density tier covers the smaller English cities and larger market towns: Cambridge (CB), Norwich (NR), Peterborough (PE), Lincoln (LN), Derby (DE), Shrewsbury (SY), Hereford (HR), Gloucester (GL), Oxford (OX), Salisbury (SP), Dorchester (DT), Exeter (EX), Plymouth (PL), Yeovil (BA), Carlisle (CA), Durham (DH), York (YO), Aberdeen (AB), Dundee (DD) and Edinburgh suburbs (FK, KY, KA, ML). Mean annual volume 5,291 tests, roughly half the suburban-steady mean. The cluster passes slightly above the UK average; the lift comes from route mix (more single-carriageway exposure, fewer multi-lane gyratories) and from the candidate pool skewing slightly older and more rural-licence-holding.

Rural easy

Rural density tier is small in centre count (20 centres) but high-passing. The cluster covers HS (Western Isles), IV (Highland), KW (Caithness and Orkney), PH (Perthshire and Highland), DG (Dumfries and Galloway), TD (Scottish Borders), SA (south Wales coast including Swansea and Pembrokeshire), LL (north Wales coast including Bangor and Wrexham), and TR (West Cornwall). Mean annual volume 2,642 tests, which is around a third of the suburban-steady mean. The cluster averages 51.5% pass rate, the highest of the five. Two mechanisms drive this: route mix (single-track A-roads, lower traffic density, fewer multi-lane junctions) and candidate-pool composition (older average age, higher private-practice hours per candidate, more dual-licence-from-tractor exposure).

The cluster top region is Wales with 10 centres. The candidate implication is straightforward: if your local centre falls in the rural-easy cluster, the cluster is doing some of the work for you. The cluster does not eliminate the need to prepare, and the cluster average masks within-cluster variation that can swing twenty points either way. But the cluster average is, on a per-centre basis, the highest the network offers.

Section 7, How to use these clusters when booking a test

The clusters are a mental model, not a prescription. Three sensible ways to use the analysis.

Identify your local cluster first. Look up your nearest centre on the pass-rate finder or the easiest centres ranking; cross-reference its postcode area against the cluster rules in the methodology block. Most candidates land in either suburban-steady or urban-mid-mixed; for them, the cluster average is roughly the UK average and the cluster framing mostly tells them not to expect outsized variance.

Decide whether to widen your search. If your local cluster is inner-city pressure and a market-town-middle or rural-easy alternative sits within an hour’s drive, the cluster difference is real enough to justify the consideration. The travel-vs-stay tradeoff is covered in the on-site guide on whether to travel for an easier test centre; in summary, the cluster gap is one of several factors but not the deciding one.

Use the cluster as a noise filter on the rankings. If your local centre’s pass rate looks anomalously high or low compared with its cluster average, that is a flag to check the centre’s sample size. Centres with under 1,500 annual tests can swing wildly year on year; the cluster average is a more stable reference point than the centre’s single-year figure.

Section 8, Methodology and limitations

Data source. DVSA DRT122A, financial year 2024-25. Each centre’s annual total tests and pass rate is read from the dataset row, postcode-area density tier is looked up from the POSTCODE_AREA_DENSITY map shared with the pass-rate-vs-population-density research page, and the cluster assignment is deterministic on the three inputs. Licensed under the Open Government Licence v3.0.

Cluster rules. Each centre is assigned by the following cascade, evaluated top-down. A centre lands in the first cluster whose rule it satisfies.

• Inner-city pressure. Urban-core density tier, 5,000 or more annual tests, pass rate below 45 percent.
• Urban mid mixed. Urban-core density tier, anything not caught by the inner-city-pressure rule (lower volume or higher pass rate).
• Suburban steady. Major-suburban density tier (typical 1,000 to 5,000 per square kilometre), any annual volume.
• Market town middle. Market-town / mixed density tier (typical 200 to 1,000 per square kilometre), any annual volume.
• Rural easy. Rural density tier (typical under 200 per square kilometre), any annual volume.

Why these cuts. The 5,000-test floor on inner-city pressure is the volume above which the network’s urban-core distribution starts to tail off; below 5,000, urban-core centres carry too much sample noise to be confidently differentiated. The 45% pass-rate threshold sits between the inner-city-pressure cluster mean (41.7%) and the urban-mid-mixed cluster mean (50.8%); it is the cleanest cut on the urban-core sub-distribution. The four density tiers are inherited unchanged from the postcode-area-to-ONS-density map used on the population-density research page.

Centres excluded. 0 car centres failed cluster assignment, almost all because the OSM display name did not carry a resolvable postcode and the name-based fallback map did not match. They concentrate in small island and Highland centres whose annual volume is already below the 500-test floor; the impact on the cluster headline figures is under 1%. Motorcycle, LGV and MPTC centres are excluded from the analysis set per the shared isCarCentre filter used across the research pages.

What this analysis does not do. The clusters are descriptive, not causal. The cluster a centre falls into is the single best one-shot predictor of its pass rate, but the prediction explains roughly 35 to 45% of the variance in per-centre pass rate; the remainder is centre-specific (route mix, examiner staffing, candidate-pool composition) and not captured by the cluster. The clustering also does not split by candidate age, by first-attempt vs retake, or by gender; those splits are covered in the dedicated research pages on each of those facets. Northern Ireland centres are not included because DRT122A does not cover them; the DVA equivalent is published separately and not yet ingested.

Method honesty. The cuts are rule-based, not the output of an unsupervised clustering algorithm. We tested the assignment against a three-axis k-means run with k=5 and the two converged on roughly 80% of centres. Disagreements were concentrated at the urban-core / major-suburban boundary, which the k-means version wanted to split on volume rather than density. We chose the rule-based version because it is interpretable (a candidate can look up their postcode area and apply the cascade themselves) and stable as new annual data lands.

Cite this page: passrates.uk research/centre-difficulty-clustering v1.0 (2026). Data: DVSA DRT122A "Driving test and theory test data, cars" for 2024-25, OGL v3.0. Aggregation: 273 GB car centres, five rule-based clusters.

Frequently asked questions

Related on PassRates.uk

• How we calculate pass rates
• Pass rate vs population density
• Centre volume vs pass rate
• London vs UK pass rate
• Wait times by region
• Pass rate by month and region
• Easiest UK test centres
• Hardest UK test centres
• Should I travel for an easier test centre?
• Pass-rate finder tool