Condo resale records: psf varies 5% to 14% inside a project, quantum up to 70%

Official transaction records show what buyers actually paid, by contract month and sale type, and they show that inside one condo project the price per square foot varies far less than the total price. In the 12 busiest resale projects, the middle half of psf spanned 5.3% to 14.0% of the median, against 31.4% to 70.2% for quantum 1.

That gap is the most useful thing a household can take from the records before making an offer. This guide is for a couple shortlisting a resale condo unit who has pulled up the project's past sales and wants to know what the numbers can be trusted to say. It covers what a URA record contains, what it leaves out, how much prices vary within one project, why a project's median price can fall while its psf rises, and how HDB's resale records differ.

Two terms first. Quantum is the total price paid. Psf is that price divided by the area URA records, converted from square metres to square feet 1. A new sale is a unit bought from the developer; a sub sale is a unit resold before the project's completion; a resale is a unit resold after it 1.

What does an official condo transaction record contain?

The Urban Redevelopment Authority (URA) publishes private residential transactions through its Data Service, covering the last five years 1. Sightlines queried it on 17 October 2026 for condominiums and apartments, leaving out landed homes and bulk purchases 1. In the 12 contract months from October 2025 to September 2026, the records hold 8,110 new sales, 646 sub sales and 11,924 resales 1.

Each record carries a contract month, a sale type, a price and an area in square metres, along with the project, its postal district, its market segment and its tenure 1. Psf is worked out from the price (URA's nett price where it gives one) and the area 1. Sightlines reports only aggregates, never a single unit's sale.

URA groups the market into three segments. The Core Central Region (CCR) is postal districts 9, 10 and 11, the Downtown Core planning area and Sentosa 2. The Rest of Central Region (RCR) is the rest of the Central Region 2. The Outside Central Region (OCR) is the East, North East, North and West planning regions 2.

Over those 12 months, the median condo resale island-wide was $1.72m, or $1,776 psf 1. By segment, the median resale was $2.68m ($2,177 psf) in the CCR, $1.81m ($1,959 psf) in the RCR and $1.501m ($1,566 psf) in the OCR 1.

Which sales are missing, and why do the latest months keep changing?

URA's records are built from caveats. A caveat is a legal document lodged with the Singapore Land Authority by a purchaser to protect their interests after an option to purchase is exercised or a sale and purchase agreement is signed, URA's glossary says 2. Lodging one costs $64.45, inclusive of GST, under SLA's fee schedule 6.

Two consequences follow, both stated by URA. Lodging a caveat is voluntary, so the transaction database may not reflect all the units sold directly by developers 2. And there is usually a time lag of a few months from the purchase of a unit direct from a developer to the lodgement of its caveat 2.

The records also keep filling after the fact. When Sightlines ran its queries, October 2026 was left out because it was still being reported 1. A buyer looking up a project can treat the most recent month or two as provisional, and new sales from developers in particular may appear months later 2.

URA's own quarterly price index uses a different and faster route. Its flash estimate for the third quarter of 2026, released on 1 October 2026, was compiled from transaction prices submitted for stamp duty and from developers' unit sales data up to mid-September 3. It put non-landed private prices up 0.9% for the quarter, with the OCR up 2.2%, the RCR up 0.2% and the CCR down 0.1% 3. URA advised the public to interpret the flash estimates with caution and said it would update them on 23 October 2026 3.

An index and a median from the raw records are built differently. The flash index draws on stamp duty submissions and developers' sales data 3. A median from the raw records tells you what the homes that happened to sell in a window went for, whatever their size, age and location.

How much do prices vary inside one condo project?

To measure this, Sightlines took every condo project with 66 or more resales in the 12 months to September 2026 1. There are 12 of them, with 1,126 resales between them 1. For each one, the table gives the median, the middle half of resales (from the 25th to the 75th percentile) and the width of that middle half as a share of the median. Sightlines calls that width the spread.

Source: URA Data Service, private residential transactions, contract months October 2025 to September 2026 1; Sightlines analysis. Spread is the 75th percentile minus the 25th, divided by the median.

The pattern is consistent. Within each of the 12 projects, the psf spread ran from 5.3% at Kingsford Waterbay to 14.0% at Jadescape and Parc Clematis, with a median of about 8% across the 12 1. Across a whole segment, the psf spread was 25.9% to 28.4%, and island-wide it was 34.2% 1. The project is doing most of the work: once you know which development a unit is in, its psf is fairly well pinned down.

Quantum is a different story. Within the same projects, the quantum spread ran from 31.4% at Stirling Residences to 70.2% at Jadescape, with a median of about 61% 1. Five of the 12 projects had a quantum spread above the 65.1% for every condo resale island-wide 1. A project's median price, in other words, can say almost as little about one unit's price as the island-wide median does.

The likely reason is unit size, though the aggregates do not give sizes, so this is an inference rather than a measured fact. If a project sells smaller and larger units side by side at a broadly similar psf, their total prices differ because their areas do, and the median quantum depends on which sizes happened to sell.

Why can a project's median price fall while its psf rises?

Comparing each project's last 12 months with the 12 months before shows the same effect over time 1. For 11 of the 12 projects, both windows have enough resales to compare; Normanton Park had fewer than 5 resales from October 2024 to September 2025, so the records give no figure for it 1.

Source: URA Data Service, private residential transactions, contract months October 2024 to September 2026 1; Sightlines analysis.

Median psf moved within a narrow band, from a 1.4% fall at Affinity at Serangoon to an 8.3% rise at Parc Clematis 1. Median quantum moved from a 26.1% fall at Sims Urban Oasis to a 7.4% rise at Parc Clematis 1.

Take Treasure at Tampines. Its median resale psf rose 2.7%, from $1,752 to $1,799, while its median resale price fell 18.7%, from $1.55m to $1.26m 1. A headline built on the median price would report a project in decline. The psf says the opposite. The most likely explanation is that smaller units made up more of the latest year's resales; the aggregate figures do not give unit sizes, so that reading is an inference, not a measured fact. At Sims Urban Oasis, psf barely moved (down 1.1%) while the median price fell by $447,500, or 26.1% 1.

Island-wide, the two measures moved more closely: the median condo resale psf rose 2.8% and the median price 4.4% between the two windows 1.

A worked example: one unit against its project's record

Suppose the couple is weighing a resale unit of 1,000 sq ft, a size chosen for round arithmetic, at Treasure at Tampines 1. It is not a specific unit.

The project's middle half of resales ran from $1,731 to $1,858 psf over the 12 months to September 2026, with a median of $1,799 psf 1. At 1,000 sq ft, the median psf works out to a quantum of $1,799,000 1. The table multiplies each figure by the same area.

Source: URA Data Service 1; Sightlines calculation.

That gives a range of $127,000 for a unit of that size 1. The project's quantum record gives a middle half of $1.14m to $1.964m, a range of $824,000, which is more than six times as wide and mostly reflects other unit sizes 1. For a buyer, the first range is a useful check on an asking price. The second barely narrows anything down.

The psf range still hides real differences between units of the same size: floor, facing, stack, layout, condition and whether it has been renovated. A unit near the top of the project's psf range needs a specific reason to sit there. A unit near the bottom may have one too. The record shows where a price sits; it cannot say why.

What do HDB's resale records show differently?

HDB's resale records, published on data.gov.sg, answer the same question for public housing but are built differently. The dataset covers resales from January 2017, organised by registration month rather than contract month 5. Its 11 columns are the month, town, flat type, block, street name, storey range, floor area in square metres, flat model, lease commencement date, remaining lease and resale price 5.

Three notes on the dataset matter for anyone using it. The approximate floor area includes any recess area purchased 5. The records exclude resales that may not reflect the full market price, such as transfers between relatives and part-share sales 5. And resale prices should be taken as indicative only, because the price agreed depends on many factors 5.

The practical differences from URA's records follow from that. An HDB price is registered in a month; a URA price is contract dated in a month, so the two are not lined up to the same point in a sale. HDB's records give the block and a storey range, which lets a buyer compare like with like inside one block; URA's aggregates, as Sightlines uses them, stop at the project. And HDB's dataset carries the remaining lease on every row 5, which matters more for older flats than it does for the young condo projects in the tables above.

When do the records mislead?

The records are a strong check, but four situations weaken them.

The first is a change in how floor area is measured. URA, SLA, the Building and Construction Authority and the Singapore Civil Defence Force harmonised their floor area definitions with effect from 1 June 2023, for development applications submitted to URA on or after that date 4. Under the revised definitions, all voids are excluded from strata area 4. A project approved under the old definitions and one approved under the new may not compare fairly on psf. Across that break, compare on quantum for a similar layout as well as on psf.

The second is a thin record. A project with few resales in a window gives a median that may rest on a handful of units. The URA tool Sightlines uses gives no figure from fewer than 5 sales and no quartile from fewer than 10 1, and a buyer working from raw records should apply a similar caution.

The third is mixing sale types. Over the same 12 months, the median new sale island-wide was $2.259m, or $2,666 psf, and the median resale $1.72m, or $1,776 psf 1. A new sale and a resale in the same district are different products, sold at different stages; compare a resale unit with resales.

The fourth is the provisional tail. The current month is still being reported 1, and for units bought direct from developers URA notes a lag of a few months before the caveat is lodged 2.

What would change the answer?

The decision rule from the records is simple. To check a resale condo price, use the project's psf range for the last 12 months, multiply it by the unit's area, and see where the asking price falls, as the Treasure at Tampines example does 1. Do not use the project's median quantum, which shifts with the mix of unit sizes that happened to sell. In 3 of the 11 projects with two years of data, the median quantum fell by more than 10% in a year while the median psf moved by less than 3% either way 1.

The counter-case is a project where psf varies with unit size. If smaller units in a project sell at a higher psf than larger ones, a small unit and a large one will sit at different ends of the range. Where the project has enough sales, compare with units of a similar size; where it does not, the project-wide psf range is a starting point, not a finding.

The rule also breaks when the market moves fast. A 12-month window averages the year. If prices moved sharply within it, the earlier months may no longer reflect what sellers will accept. URA's next update of its third-quarter index, due on 23 October 2026, is one way to see whether the whole market has shifted since mid-September 3.

Not making an offer can be a valid answer too. An asking price well above a project's recent psf range, with no reason given for it, is a price the record does not support.

This is general information, not financial advice. Check the current rules with IRAS, HDB, CPF Board or your bank before you commit.