Land Registry records 855,564 flats in England and Wales that currently carry a valid Energy Performance Certificate. Of those, 57,499 — 6.7%, or roughly one in fifteen — have an EPC record that classifies the property not as a flat or maisonette, but as a house. The two national registers that underpin most of the UK's property data disagree with each other on a basic question — what kind of home is this — more often than most buyers would expect.
The gap is not random. It tracks the age of the building closely, and it barely shows up at all when the same check is run in reverse on houses.
How the two registers can disagree
HM Land Registry's Price Paid Data assigns every sale one of five property types — detached, semi-detached, terraced, flat/maisonette, or other — based on how the transfer was registered. The separate EPC register, run under different rules by a different body, records its own property type when an energy assessor certifies the building, using categories including house, bungalow, flat, maisonette and park home.
Most of the time the two agree. But a building's legal form at the Land Registry and its physical form as an assessor sees it on the day of inspection are not always the same thing. A large Victorian or Georgian house legally subdivided into separate flats will show up as "flat" at the Land Registry the moment each unit is sold off individually — but an EPC assessor visiting one of those units, especially if floor plans or internal layout still read as "part of a house" rather than a purpose-built block, can still log the property type as "house". The reverse can happen too: terraces and semis whose internal layout has been split or converted are sometimes logged as flats or maisonettes by an assessor even though Land Registry still carries them as a house-type sale record.
The headline numbers
Taking the latest EPC on file for every address (the standard method for any current-state EPC figure, since older certificates on the same property are superseded), the national cross-tabulation for Land Registry-classified flats looks like this:
| EPC register says | Count | Share of LR flats |
|---|---|---|
| Flat | 676,492 | 79.1% |
| Maisonette | 116,502 | 13.6% |
| House | 57,499 | 6.7% |
| Bungalow | 5,067 | 0.6% |
Source: HM Land Registry Price Paid Data joined to the EPC register by address, latest certificate per address, England & Wales, fetched 21 October 2026.
Older buildings disagree twice as often
Splitting the "flat" population by the building's recorded construction era shows the mismatch is far from evenly spread:
| Build era | LR flats with an EPC | Recorded as "House" | Disagreement rate |
|---|---|---|---|
| Before 1930 | 134,882 | 14,409 | 10.7% |
| 1930–2002 | 509,920 | 29,665 | 5.8% |
| 2003 onwards | 157,286 | 7,981 | 5.1% |
A flat in a pre-1930 building is more than twice as likely to carry a "house"-typed EPC as a flat built since 2003. The pattern fits the conversion story: pre-1930 stock includes the bulk of England and Wales's subdivided townhouses and villas, while post-2003 flats are overwhelmingly purpose-built blocks that were never anything other than flats to begin with — leaving far less room for an assessor's classification to drift from Land Registry's.
The reverse check — and why it's weaker
If grand old houses carved into flats explain the gap above, the same logic should show up in reverse: older houses being logged as flats or maisonettes more often than newer ones. Running the same test on Land Registry's terraced-house population — the type most associated with subdivision — finds a much flatter pattern:
| Build era | LR terraces with an EPC | Recorded as Flat/Maisonette | Rate |
|---|---|---|---|
| Before 1930 | 1,470,660 | 35,664 | 2.4% |
| 1930–2002 | 1,549,951 | 35,071 | 2.3% |
| 2003 onwards | 224,205 | 4,908 | 2.2% |
Era barely moves the needle here. What does move it is the type of house itself. Across all property types nationally, detached houses are rarely reclassified as flats (0.45%), semi-detached houses more often (1.08%), and terraced houses most of all (2.28%) — a five-fold spread that tracks the housing form, not its age. Terraces are simply the house type most often split into separate dwellings after being built as one, regardless of when that building went up.
It doesn't change your council tax band
This is a data-quality quirk between two separate registers, not a legal reclassification of the property. Council Tax bands are set independently by the Valuation Office Agency based on 1991 (England) or 2003 (Wales) capital values, and are unaffected by how a later EPC assessor happened to categorise the building type. A flat with a "house"-typed EPC on file is still a flat for every practical purpose — council tax, mortgage lending, leasehold status — that matters to an owner or buyer. The mismatch mainly matters for anyone doing research that depends on EPC property-type fields being accurate, including, at times, this site's own area-level EPC analysis.
A concrete example
Gregory Place, a short Georgian terrace in Kensington (W8), illustrates the pattern cleanly: Land Registry carries several of its subdivided units as flat sales, while EPC certificates for the same block mix "Flat" and "House" property types depending on which unit an assessor logged. See the full property history for W8 4NG on Homecost's postcode tool.
Methodology
Figures use HM Land Registry Price Paid Data (category A, standard price-paid entries) matched to EPC Open Data certificates by address, taking the most recent certificate on file for each address where more than one exists. Build era uses the EPC register's own age-band field, bucketed as before 1930, 1930–2002, and 2003 onwards; rows with no recorded age band are included in national totals but excluded from the era breakdown. All figures were queried on 21 October 2026 and cover England and Wales, the geographic scope of both registers.
This finding grew out of an earlier look at EPC ratings across prime central London flats, where period-conversion stock first showed up as an outlier worth checking nationally. It pairs with Homecost's wider work on where EPC certificate coverage is thinnest — together the two pieces are a reminder that the EPC register, useful as it is, has real gaps and inconsistencies that are worth checking before leaning on it too heavily for any single address.
Browse more cost intelligence guides on Homecost.