How current the data is
| Sale records | 813,106 |
|---|---|
| Latest sale on the register | 16 October 2026 |
| Register last checked | 21 Oct 2026 |
| Statistics rebuilt | 21 Oct 2026 |
| Source file SHA-256 | a5283913324a8b63… |
Sales reach the register when stamp duty is filed, often weeks or months after the sale. So the latest quarter and year are always incomplete, and charts mark them that way.
Rules for every statistic
- Sales marked “not full market price” are left out of every median, percentile and index.
- Bulk sales (five or more homes in one county registered on the same day at the same price) are left out, because their prices are usually a total split across the homes.
- Prices published without VAT (almost always new homes) have 13.5% VAT added, so they compare fairly with second-hand sales. Sale pages always show the published price.
- Medians are hidden for groups with fewer than 10 sales, and groups with fewer than 30 are marked as small samples.
- Figures are not adjusted for inflation or, unless stated, for the mix of homes sold.
- Records the PSRA later corrects or withdraws are left out of statistics, but their pages stay available.
New and second-hand prices on the same footing
The register publishes new-home prices without VAT and second-hand prices with it. Compared directly, new homes would look about 14% cheaper than they are. For medians and ranges we add VAT at 13.5% to prices published without it. Sale pages always show the price exactly as published, with the VAT-inclusive figure beside it.
Sample sizes
A median of a handful of sales is mostly noise. Groups with fewer than 10 market-price sales show no median. Groups with fewer than 30 are drawn hollow on charts and marked “small sample”. Every figure comes with the number of sales behind it.
Areas
The register has no coordinates, and only about a third of records have an Eircode. Areas come from the published address text:
- County: as published.
- Dublin postal district: from the Eircode routing key (D01–D24, D6W) or, failing that, “Dublin N” in the address.
- Eircode area: the three-character routing key, for records with a valid Eircode only. Coverage is partial and uneven, so these figures describe the sales that have an Eircode, not every sale in the area.
- Town: address components matched against OpenStreetMap place names within the same county.
Address parsing is rule-based text matching. It can misplace an unusual address within its county, but it never moves a sale to a different county.
Matching sales of the same home
Sales of the same home are grouped by tidying the address: case, punctuation, common abbreviations such as “Rd” for “Road”, and county suffixes. Two sales count as the same home only if the tidied addresses match exactly within one county, and the address has a house number or name. This misses some genuine matches, such as a house name added later, rather than risk joining two different homes.
Links and corrections
Each register row gets a stable identifier made from its published fields, so links keep working across register releases. When the PSRA corrects or withdraws a record, the old version is marked as removed. Its page stays up, but it is left out of every statistic.
The Property Price Index (repeat sales)
Medians move when the mix of homes sold changes. The index instead uses homes sold more than once: each pair of consecutive market-price sales of the same home is one observation of the price change between two dates. A regression of log price ratios on period dummies (Bailey–Muth–Nourse) combines the pairs. Case–Shiller weights give less weight to pairs far apart in time, whose changes carry more noise from renovation and wear. The first period is set to 100, and 95% confidence intervals come from the weighted regression.
- Pairs must be at least 6 months apart, change by less than 3× in either direction, and by no more than 50% a year.
- Bulk, multi-unit and “not full market price” sales are left out. New-home prices include VAT.
- Quarterly series need 4,000 pairs, annual series 400. Periods with fewer than 15 pairs, or not linked to the base period by any chain of pairs, are left blank rather than guessed.
- The index is checked against the CSO Residential Property Price Index, a hedonic index from stamp-duty and BER data. Repeat-sales indices usually rise a little more slowly over long spans because homes age between sales.
- Each series is revised when new register data arrives, because later sales add pairs to earlier periods.
The 2007 peak and inflation
The register starts in 2010, after the crash, so comparisons with the 2007 peak use the CSO Residential Property Price Index (table HPM09), which runs monthly from 2005 and is adjusted for the mix of homes sold. The peak is the highest month before the post-crash low, for Ireland, Dublin and outside Dublin.
“After inflation” figures restate prices in the euros of the latest month, using the Eurostat Harmonised Index of Consumer Prices for Ireland (all items). A price from a past year is multiplied by the latest month’s index divided by that year’s average index. Before-inflation figures are the default everywhere.
Places and map positions
Town names and approximate positions come from OpenStreetMap place names (cities, towns, villages, suburbs, townlands) fetched for each county. An address is matched only to places in its own county; a name shared by places more than 5 km apart is ignored unless a town in the same address resolves it. Map circles sit at the median position of an area’s matched sales. No individual home is placed on a map.
Data-quality signals
Signals describe patterns in the published data. They never change or remove a record.
| Part of a bulk transaction | Five or more homes in the same county were registered on the same day at exactly the same price. This usually means a portfolio or block sale with the total apportioned across units, so the price may not reflect an individual home’s value. |
|---|---|
| Possible duplicate | Another register record has the same normalized address and sale date. |
| Multiple units in one record | The address appears to cover more than one unit (for example “4 & 5 Main Street”), so the price may be for several homes. |
| Unusually low price | Below €30,000 without a “not full market price” flag. This can be a share of a property, a site, or a data entry issue. |
| Unusually high price | Above €5 million. Genuine for some homes, but also typical of multi-unit or mixed-use transactions. |
| Large change since previous sale | The price changed by more than 3× (up or down) since this home’s previous recorded sale, or by more than 50% a year. Possible causes include major works, a different unit at a similar address, or an error. |
Other sources
- CSO Residential Property Price Index (table HPM09) and RTB Average Monthly Rent Report (table RIA02), © Central Statistics Office, CC BY 4.0.
- Eurostat house price index (prc_hpi_q) and Harmonised Index of Consumer Prices (prc_hicp_minr), © European Union, Eurostat.
- Place names © OpenStreetMap contributors, ODbL. County boundaries © Tailte Éireann, CC BY 4.0; Northern Ireland outline from Natural Earth (public domain).
- Mortgage limits from the Central Bank of Ireland mortgage measures; dated on the affordability page.
- Planning permissions: National Planning Application database (Department of Housing), fetched only as per-council yearly totals; applicant names and addresses in that dataset are never downloaded.
What this site is not
Nothing here is a valuation of any property. Figures are not adjusted for inflation and, apart from the index, not adjusted for the mix of homes sold. The register itself may contain errors, and we publish the data as received.