Road condition

The unclassified road blind spot

Unclassified roads are about 62% of England's local network and the least systematically surveyed. What the data actually covers, and why.

Most of the road network in England is the part nobody measures properly. This page sets out how the classification works, what condition data exists for each class, and where the gap sits.

How local roads are classified

English roads are classified as A, B, C or unclassified. A roads provide large-scale links within and between areas; B roads connect areas and feed traffic between A roads and smaller roads; C roads (formally, classified unnumbered roads) link estates and villages into the network; unclassified roads are the smallest roads, used for local traffic. Since April 2012 the classification decision sits with the local highway authority rather than requiring approval from the Secretary of State, and DfT guidance is explicit that classification is relative to local conditions, not to a national minimum standard (DfT, Guidance on road classification and the primary route network).

How much of the network this is

The local road network in England is around 183,000 miles, or 98% of all road length. Using DfT's 2022 road length data, the National Audit Office breaks it down as A roads 17,900 miles (9.8%), B roads 12,400 (6.8%), C roads 39,700 (21.7%) and unclassified roads 112,600 (61.7%) (NAO, The condition and maintenance of local roads in England, July 2024, Figure 1). DfT's own figure is consistent: unclassified roads accounted for around 62% of the local authority managed network in 2024 (DfT, Road conditions in England to March 2025, published 21 January 2026).

Worth knowing: DfT's headline road length series no longer separates C from U in most recent years. Table RDL0103 records C and U combined for 2023, 2024 and 2025 because no R199b road length consultation with authorities took place; 2022 is the most recent year with a clean split (DfT, road length statistics RDL0103, updated 12 February 2026).

These roads are not lightly used. Local A roads and locally managed motorways are about 10% of the local network and carry roughly 31% of traffic. Local authority minor roads — B, C and unclassified together — carried 99.8 billion vehicle miles in England in 2024, about 35% of all motor traffic, slightly more than the entire strategic road network (DfT table RDC0301). And as the NAO puts it, almost all journeys, by any mode, start and end on the local road network.

How condition is surveyed, class by class

Classified roads. Authorities have been required to use SCANNER (Surface Condition Assessment for the National Network of Roads) — accredited survey vehicles measuring rutting, texture, cracking and profile. For A roads and locally managed motorways the requirement is 90% of the network in both directions over two years, so the 2025 figures were collected between April 2023 and March 2025.

Unclassified roads. There is no mandated method. DfT states plainly: "There is no mandated method for the collection of condition data on unclassified roads, resulting in a variety of road condition monitoring techniques used by local authorities. CVI and DVI surveys have been the most frequently used survey types, but these have been decreasing in recent years" (DfT table RDC0131, note 2). A Coarse Visual Inspection is a rapid assessment, usually from a slow-moving vehicle; a Detailed Visual Inspection is walked and records more defect classes, and is typically targeted at lengths already known to be defective (DfT technical note, section 3).

The coverage expectation is also different: 90% of the unclassified network over four years, and expected rather than required. The unclassified figures published in January 2026 were collected between April 2021 and March 2025.

What the statistics actually report

For the year to March 2025: 5% of local A road, 7% of B and C road and 17% of unclassified road should have been considered for maintenance. Do not read that as unclassified roads being three times worse. DfT is direct about this — condition is scored against different definitions for different classes, so "the percentage of red road is not comparable across road classes". The unclassified figure has moved between 15% and 18% since reporting began in 2008.

Nor is it a full picture. In 2025, 153 authorities returned condition data; 149 included A road figures and 133 included B and C. Counting the published table RDC0130, around 130 authorities carry an unclassified figure for the year to March 2025. Regional and national unclassified figures only include the authorities that supplied data that year. London shows how fragile this is: after Transport for London withdrew funding for classified road surveys, only 12 of 33 boroughs provided valid unclassified data in 2020, and DfT suppressed the London figures for three years.

The NAO's assessment in 2024 was that there is no reporting requirement for unclassified roads at all, and that DfT limited what it collects to minimise administrative burden.

Where the method mix is going

Compliance with the SCANNER mandate has been falling. In 2025, 79% of authorities submitting valid data used SCANNER on A roads and 65% on B and C roads, down from 99% and 98% in 2018. Authorities have been adopting other technologies faster than the reporting regime recognised them.

PAS 2161:2024, published in September 2024, is the response. It replaces the single technology mandate with a list of approved technologies, and replaces red/amber/green with five categories based on the treatment needed. DfT published the approved technology list in September 2025, valid to March 2027; authorities must collect data using an approved survey for the financial year April 2026 to March 2027, with full transition by the year ending March 2028, and there will be a break in the statistical time series (DfT, PAS 2161 guidance).

One detail matters here. The standard itself is written to cover locally managed classified and unclassified roads (TRL), but DfT's national reporting requirement under it applies to classified local roads — Single Data List items 130-01 and 130-02. The reform opens the market. It does not, on its own, put unclassified roads inside the mandate.

Why this is an operational problem, not a statistical one

Three consequences follow, and it is worth being precise about each.

Funding allocation does not depend on condition. DfT allocates almost all capital maintenance funding by formula: 82.4% on road length, 15.4% on bridges, 2.2% on lighting columns. The NAO found DfT "does not use data on condition — which are partial and incomplete — to identify areas in greatest need". So the argument for better unclassified data is not that it unlocks more grant. It is that it determines where your own programme goes.

Accountability does depend on it. From 2025-26 DfT introduced transparency reporting, annual ratings and incentive funding. For 2026-27, publishing a transparency report unlocks 50% of the incentive element, with further tranches for evidence of spend, an updated asset management policy and strategy, and continued professional learning (DfT letter on 2026-27 funding). What you can evidence about your network is now visible.

The public experience sits on the unmeasured part. DfT itself notes that user perception "may be heavily influenced by the condition of the unclassified road network". The Asphalt Industry Alliance's ALARM 2026 survey reports a record £18.62 billion carriageway repair backlog, an average carriageway budget shortfall of £8.1 million per authority, and £31.0 million spent across England, London and Wales on compensation claims and the staff time to handle them, 99% of which relate to potholes (ALARM 2026, published 17 March 2026). No national breakdown of those claims by road class appears to be published — which is itself part of the point.

The constraint authorities describe is not indifference. It is that a designed survey of 112,600 miles of residential street, at the frequency that would make it useful, costs more than the budget allows. DfT's stated reason for lower unclassified coverage is the size and relatively lower priority of the network; the NAO's finding is that data collection was limited to reduce burden. Both are cost arguments.

Where Make Sense fits, and where it does not

Make Sense is building a multi-sensor pod designed to mount to vehicles already making their rounds — refuse collection, social care, maintenance fleets — and to turn those journeys into continuous observation of road surface condition alongside air quality and building heat loss.

The proposition is about the economics of collection, not about surveying. If the vehicle, the driver and the route are already funded, the marginal cost of observing the surface changes shape. Refuse rounds in particular touch nearly every residential street on a weekly cycle — the same streets that sit outside the classified survey programme.

The honest bounds:

The first vehicle system is in build. Nothing is deployed and nothing is operating.
Make Sense is not PAS 2161 approved. Data from a system that is not approved cannot be used for national reporting of road condition under the requirements taking effect for April 2026 to March 2027.
There are no published accuracy or performance figures, because nothing has been validated.
This page makes no claim that Make Sense has surveyed any road.

What riding on existing rounds does not give you

This matters as much as the upside. Capture on operational vehicles covers where those vehicles go, when they go there. It is not a designed survey route.

Refuse rounds do not run the length of every rural C road. Fleet schedules change with rounds, seasons and service redesign. Frequency is uneven by construction: some streets are passed weekly, others rarely, some not at all. Speed, lane position and time of day are set by the operational task, not by survey design. And a network where coverage is a by-product of service delivery will have gaps that correlate with service patterns rather than with engineering need.

That means opportunistic capture is a complement to a survey programme, not a replacement for one. Anyone claiming otherwise is selling.

If this is your problem

If you are responsible for asset management on a network where the unclassified majority is surveyed on a four-year expectation and everything else is inference, we would rather hear how you would want that tested than tell you what we think. Get in touch.

Sources

Last reviewed 12 August 2026