How to read this data
1. “Has been under an order” is not “is under an order”
The Buildings Department publishes orders issued and orders complied with / withdrawn / superseded as two separate datasets, with different start dates, and never publishes which orders are still outstanding.
A building appearing on the issued list tells you an order was issued at some point. It does not tell you the order is still open — the work may have been done years ago. Nor does the cleared list settle it: the two datasets do not share a key that lets you match one record to the other reliably.
This is why every page on this site says “on the Buildings Department’s list of orders issued” and never “has outstanding orders”. If you need the current status of a specific order, ask the Buildings Department.
Our daily archive since 2026-08-31 is the closest available evidence: a record that leaves the issued list between two snapshots has most likely been dealt with. That is an inference from two observations, not a government statement, and we label it as such.
2. Unauthorised building works only go back to 1 May 2023
The removal-order dataset starts on 1 May 2023. A building with no unauthorised-building-works order on this site may still have unauthorised structures — the order may predate the dataset, or no order may have been issued yet. No order is not a clean bill of health.
3. Age thresholds are law, not opinion
- 30 years — a private building enters the pool for the Mandatory Building Inspection Scheme and can be selected at any time. Banks also shorten mortgage terms on older buildings, which changes what a buyer can borrow.
- 50 years — from 6 December 2024, the compulsory sale (強拍, forced auction) threshold drops to 70% ownership in designated districts, 80% elsewhere, for buildings aged 50–59. The older the building, the lower the threshold.
Age on this site is measured from the occupation permit of the oldest block in the building. A redeveloped tower on an old podium can read older than it looks.
4. Storey counts in the government label are often wrong
The storey figure published with building records disagrees with the building itself often enough that we do not rely on it — one well-known tower is labelled 22 storeys and has roughly 47. Where we have a 3D model we count floors from the model geometry instead.
An unusually large metres-per-floor figure is usually real, not an error: buildings skip floor numbers for superstition, and pool decks, plant rooms and double-height lobbies genuinely take more height than a residential floor.
5. Height above ground, not height above sea level
Government building data carries a top elevation measured from principal datum — sea level — so a three-storey house on a hillside can carry a figure of 52 metres. What we show is height above the ground at the building, which is the number people mean when they ask how tall it is.
6. Not every building has a 3D model
The Lands Department 3D model covers roughly three quarters of buildings. Where a model is missing we do not substitute the nearest one — matching a building to the wrong model gets it wrong about one time in five, and a wrong model is worse than none.
7. Building height limits come from plans of different ages
Statutory building height limits are cut from outline zoning plans, and those plans were made at different times — in roughly half of them the height limit layer is older than the current plan. Treat the limit as an indication of what the plan intends, and check the plan itself before relying on it.
8. Records at an address are matched by location
Licences, lift notices, planning applications and owners’ corporation entries are matched to a building when the government coordinate falls inside that building’s footprint (±6 m). They are not matched by file number, because the datasets do not share one. An estate registered at a different address point can therefore appear to have no owners’ corporation when it has one.
9. Market figures are district averages
Rating and Valuation Department rents and prices are averages for a region and a flat size class. They say nothing about a specific flat. Within a single block, price per square foot varies by about 30% — floor level explains under a tenth of that, and two flats on the same floor still differ. Any figure quoted at building level hides that spread.
What to do with all this
Use these pages to know what to ask about, then confirm through the official channel before you act. The data sources page links every dataset so you can look up the original record yourself, and each block on a building page carries the department and the date we retrieved it.