The cheapest 5 room HDB flat sold in Singapore over July and August 2026 went for $420,000. The dearest went for $1,720,000. Same official flat type, same three letters on the same government form, and a gap of $1.3 million between them.

That range is why almost every article you will read on this topic is useless. They quote a national median, they quote a town table lifted off a dashboard, and they stop. So I did the boring thing instead. I pulled every five-room resale registration for July and August 2026 across the twelve biggest towns in the country — 831 transactions — reconciled each town’s count against the registry’s own reported total, and then went looking for the places where the numbers disagree with the story everyone tells.

They disagree in four places. The town table is less stable than anyone admits. The size range hiding under one label is enormous. The trick that explains most of the four-room town gap does not explain the five-room one. And the “natural next step” from a five-room to an Executive flat barely exists as a real market at all.

What a 5 room HDB flat actually costs, town by town

Here is August 2026. Every cell is every transaction the registry recorded for that town and flat type in that month — not a sample, not a rolling average, not a listing price. Medians, because means get destroyed by one penthouse.

A few things worth pausing on before we go further.

The spread is 1.90 times on price — Queenstown’s median is $569,000 above Jurong West’s. On psf it is 1.78 times. That is a big number, and the obvious reading is “location”. Hold that thought; section five takes it apart.

The expensive towns barely trade. Queenstown recorded seven five-room sales in the month. Ang Mo Kio nine. Toa Payoh and Bukit Merah fifteen each. Meanwhile Sengkang did 61, Tampines 56, Woodlands 51 and Punggol 50. So the four dearest towns in that table are resting on 46 transactions between them, and the eight cheaper ones on 333. Any month-to-month movement you see at the top of the table is mostly noise, and I will prove that in a moment rather than asserting it.

The median floor area runs the wrong way. Look at the sqm column against the price column. Pasir Ris has the biggest median five-room in the table at 123 sqm and sits ninth on psf. Punggol and Sengkang have the smallest at 111–112 sqm and sit sixth and ninth on price. The dear towns are not selling bigger flats — several of them are selling smaller ones at higher rates. That is a vintage signature, not a space signature, and it is the thread that runs through this whole article.

If you want to know what those medians mean for your own cheque rather than in the abstract, the stamp duty alone separates them by a useful amount. At current buyer’s stamp duty rates, Bedok’s $770,000 median attracts roughly $17,700 and Queenstown’s $1,200,000 attracts roughly $32,600 — my arithmetic, and worth checking against the free calculators at listings.sg/tools alongside your loan-to-value and total debt servicing limits before you fall in love with a postcode.

The ranking moved four places in thirty days

Every property site in Singapore publishes a town table like the one above and presents it as a fact about the world. It is not. It is a fact about one month’s transactions, and when a town only does seven or nine sales, one month is a coin toss.

So here is the same table for July 2026, computed identically, and then the two set against each other.

One town out of twelve held its rank. Pasir Ris, ninth both months. Everything else moved, four towns moved two places or more, and Bukit Merah went from fourth to first on the back of fifteen transactions.

Queenstown is the clearest illustration. In July its median five-room psf was $1,191 on nine sales. In August it was $820 on seven. Nothing happened to Queenstown in thirty days. What happened is that July’s nine sales had a median lease commencement of 2013 and August’s seven had a median of 1995 — the town sold new flats one month and old flats the next, and the “price per square foot in Queenstown” moved by nearly a third as a result.

I want to be fair to the eight towns with real volume, because they behave. Woodlands moved 0.3%. Punggol moved 0.05%. Sengkang 3.6%, Yishun 3.1%, Pasir Ris 3.9%. Where there are fifty or sixty sales a month, the median is stable to within a few per cent and the table means something. It is specifically the mature, central, expensive towns — the ones buyers most want a number for — where the number is least trustworthy. That is an awkward finding and I have not seen anyone else publish it.

Ninety-nine square metres to 149, all called a 5-room

Pool the two months and you get 831 five-room sales. Their floor areas look like this.

Fifty square metres — 538 sq ft — separates the smallest five-room sold from the largest. That is not a rounding difference. It is the size of a whole two-room flat, dropped inside a single official label.

Thirty-seven distinct floor areas appear in those 831 sales. But they are not spread evenly, and the shape is the interesting part:

So the five-room is really two products. If you are comparing a 121 sqm flat in Yishun against a 113 sqm flat in Punggol, you are not comparing like with like on space or on age, and the eight square metres you gain in Yishun cost you roughly nothing per foot while the lease you give up costs you a great deal. The ninety-nine square metre minimum, incidentally, is a Premium Apartment at Blk 92 Dawson Road in Queenstown that sold for $1,510,000 in August — the smallest five-room in the sample and one of the dearest. The 149 sqm maximum is a Premium Apartment Loft at Blk 268B Punggol Field, $1,288,000 in July. Same flat type, 50 sqm apart, and the smaller one cost $222,000 more.

The vintage test, and why it fails differently here

Last week I published a piece on how HDB flats are actually valued and, separately, worked through what price per square foot does and does not tell you about a four-room. The headline finding there was strong: hold lease vintage constant, and roughly two-thirds of the gap between the dearest and cheapest town evaporates. The town table turns out to be, in large part, a lease-age table wearing a geography costume.

I expected the same thing here. It is not what happened, and the honest thing is to publish that rather than quietly reuse last week’s conclusion.

First, the five-room sorted by the decade its lease commenced, pooled across both months and all twelve towns:

The 1970s row is above the 1980s and 1990s rows, which looks like a broken gradient until you see which towns are in it: Bedok, Bukit Merah and Queenstown dominate the thirty-five sales. It is a town-mix artefact inside a vintage cut — the exact mistake this whole exercise is meant to avoid, showing up in my own table. I am leaving it visible rather than dropping the row, because it is a useful reminder that a control in one dimension can leak in another.

Now the test. The 2010s band is the only one with all twelve towns present, so it is the best like-for-like comparison available:

Compare the ends. All-vintage, the twelve-town five-room psf spread is Queenstown $1,148 against Jurong West $498 — 2.31 times. Within the 2010s band it is Queenstown $1,296 against Woodlands $602 — 2.15 times. The compression is almost nothing. Eight of twelve towns move by one place or none, and the four central towns that top the raw table still top the controlled one.

For the four-room the same exercise squeezed the spread from 2.10 times down to 1.43. For the five-room it goes from 2.31 to 2.15. Those are different answers to the same question, and I think the reason is the most useful thing in this article.

One sanity check before I move on, because a single band could mislead. In the 2000s band (ten towns, 217 sales) the order is Queenstown, Bukit Merah, Tampines, Ang Mo Kio at the top and Jurong West, Woodlands at the bottom. In the 1990s band (ten towns, 208 sales) it is Queenstown and Toa Payoh at the top, Woodlands and Jurong West at the bottom. Three different vintage bands, the same three towns at the bottom every time. Whatever is going on at the top of the table, the bottom of it is robust: Jurong West, Woodlands and Yishun are the cheapest places to buy a five-room in Singapore at any age of lease.

The highest HDB price ever was set this month

It is a new Central Area town record. The previous mark was $1,720,000, which now sits at number three.

The national record is now $90,000 away. Bukit Merah’s $1,728,000 from April 2026 still stands. If you follow million-dollar HDB flats at all, the next Pinnacle-area transaction is worth watching closely.

The cheapest 5-room in the country

The other end deserves as much attention, because it is where the actual opportunity is and nobody writes about it.

The lowest five-room price in my 831-sale sample was $420,000: Blk 198 Boon Lay Drive, Jurong West, 112 sqm, first to third storey, Standard model, lease commenced 1976, sold July 2026. That works out to about $348 per square foot.

Set that against the $1,493 psf at Cantonment Road and you get a 4.29 times spread inside one flat type in one country in one two-month window. For $420,000 you are buying 1,205 sq ft of floor area with roughly forty-nine years of lease left on it — which is a genuinely good deal for a cash buyer in their sixties and a genuinely bad one for a thirty-two-year-old with a CPF loan, for reasons I have written about at length elsewhere: the remaining lease has to cover the youngest buyer to age ninety-five before CPF and loan tenure work normally, and at forty-nine years it does not.

This is the recurring shape of the HDB market and it is worth stating plainly. The cheap flats are cheap because of the lease, and the lease restriction bites hardest on the buyers who most need a cheap flat. That is not a market failure anyone is about to fix. It is, however, a real arbitrage for the right buyer — and the right buyer is older and paying cash.

The step up to an Executive is not a ladder

Every agent in Singapore, me included, has at some point drawn the ladder for a client: three-room, four-room, five-room, Executive. Bigger, dearer, one rung at a time. I wanted to price that top rung properly, so I pulled every Executive sale in the same twelve towns for the same two months — 275 of them — and went looking for same-block pairs, which is the only fair way to compare two flat types without town and vintage contaminating the answer.

Across 831 five-room sales and 275 Executive sales, in twelve towns, over two months, I found three blocks where both types sold.

Three pairs is not a dataset. I am publishing it anyway, because the absence is the finding. The five-room and the Executive were largely built into different blocks, so they almost never transact out of the same address, so there is no clean same-block premium to quote. I checked this by hand rather than trusting the join: there are plenty of near-misses on the same street — 716 and 633 and 721 Bedok Reservoir Road, 894 and 889 Tampines Street 81, five separate blocks on Pasir Ris Street 51 — but a different block on the same street is a different building with a different age and a different layout, and I will not pretend otherwise to manufacture a bigger sample.

The Punggol pair is the one clean control in the set: same block, same month, same storey band, same lease year, same flat model. Eighteen extra square metres for $168,000, which is +25.8% of price for +16.4% of space. Make of one observation what you will — I would not build a strategy on it, and I would not let anyone quote it at me as a rule either.

What you can say with confidence is the town-level step, because the sample sizes are real in four towns:

Punggol’s +3.2% is the outlier that explains the whole table. Punggol and Sengkang “Executives” are 2000s Premium Apartments and Apartments of 126–134 sqm, not the 146–150 sqm maisonette stock that the mature towns are selling. In the four towns with real Executive volume the step is 18% to 42% of price for 24 to 30 extra square metres. In Punggol it is 3% for 16, because it is barely a different flat. If you are weighing the jump, my honest take on executive maisonette prices goes into what you are actually buying at that end, and the jumbo flat is the stranger, cheaper cousin worth knowing about.

The town that sold no Executive flats at all

One cell in that Executive table is empty, and I checked it twice with a direct query before believing it: Bukit Merah recorded zero Executive resales in August 2026. Not one, in the town that holds the national price record.

That is not a market gone quiet. Bukit Merah did fifteen five-room sales the same month at a median of $1,095,000. It is a stock fact — the town was largely redeveloped with flats that top out at five-room, and its Executive inventory is tiny. Nationally there were 170 Executive resales in August against 578 five-rooms, a ratio of 3.4 to one, and the Executive is now 6.7% of the whole resale market. HDB stopped building them. The stock only shrinks from here, and it concentrates in the towns that were building big flats in the late 1980s and 1990s — Woodlands, Pasir Ris, Tampines, Jurong West — which is exactly where that table shows the volume sitting.

Who should buy a 5 room HDB flat, and who shouldn’t

What I would actually do

If a client came to me tomorrow with $800,000 and a brief of “best five-room I can get”, here is the order I would work in, and it is deliberately not the order most people use.

First, fix the lease before the postcode. Decide what remaining lease you need for your own age and financing, and treat it as a hard filter. This eliminates more bad purchases than any other single step, and it is the thing the town tables cannot help you with because they average over it.

Second, pick towns by volume, not by price. In a town doing fifty-plus five-room sales a month you can price a flat to within a few per cent and you can negotiate against real evidence. In a town doing nine, you are guessing and so is the seller, and the valuation is more likely to come in somewhere awkward. That matters directly for your cash-over-valuation exposure.

Third, measure the flat. Not the label. A 110 sqm and a 123 sqm are both five-rooms, and thirteen square metres is a bedroom’s worth of difference at no difference in name.

Fourth, do the arithmetic before the viewing, not after. Stamp duty, the loan-to-value limit against your actual lease, total debt servicing, the renovation number for a 1980s flat which is not the same as for a 2016 one. The free calculators at listings.sg/tools will get you most of the way in about ten minutes, and it is much easier to walk away from a flat you have not already emotionally bought.

Fifth, and only fifth, compare against a condo. A $1.1 million five-room in a mature town and a small condo are genuinely competing purchases at that number, and the honest comparison is not obvious in either direction. I worked through it in detail in the HDB versus condo price gap.

Frequently asked questions

What is the average price of a 5 room HDB flat in Singapore in 2026?

There is no single useful average, because the range is too wide to average honestly. In August 2026 the median five-room resale price ran from $631,000 in Jurong West to $1,200,000 in Queenstown across the twelve largest towns. The cheapest five-room sold over July and August was $420,000 in Jurong West and the dearest was $1,720,000 at Cantonment Road. Use your own town’s median and your own block’s sales, not a national figure.

How big is a 5 room HDB flat?

Across 831 sales in July and August 2026 the median was 115 sqm, about 1,238 sq ft. But the range was 99 to 149 sqm — 1.51 times — across 37 distinct floor areas. The distribution has two peaks: a modern cluster at 110–113 sqm, which is 43.1% of all sales, and an older cluster at 120–123 sqm, which is 26.1%. Always check the actual floor area on the listing rather than assuming a standard size.

Is a 5 room HDB flat cheaper per square foot than a 4-room?

Not reliably, no. This is one of the most repeated claims in Singapore property writing and it does not survive a fair test. When you compare a four-room and a five-room that sold in the same block in the same period — which removes town and lease-age effects — the five-room’s price per square foot is essentially the same as the four-room’s, and in about half of blocks it is higher. The apparent discount in national figures is a town-and-vintage mix effect, not a size effect.

Which town has the cheapest 5 room HDB flats?

Jurong West, Woodlands and Yishun, and this is the one part of the town ranking that is robust. Those three towns sat at the bottom of the price-per-square-foot table in August 2026, in July 2026, and in every lease-vintage band I tested separately — 1990s, 2000s and 2010s stock alike. August medians: Jurong West $631,000, Woodlands $650,000, Yishun $664,000.

What is the highest price ever paid for a 5 room HDB flat?

$1,728,000, for a 113 sqm flat on the 46th to 48th storey of Blk 96A Henderson Road in Bukit Merah, in April 2026. That is also the highest price ever paid for any HDB flat. Second is $1,720,000 at Blk 1C Cantonment Road in the Central Area, registered in September 2026 — which is a new Central Area record and, at $1,493 per square foot, a higher rate per foot than the national record holder despite a lower total price.

Should I buy a 5-room or an Executive flat?

In towns with real Executive volume — Woodlands, Pasir Ris, Tampines, Jurong West — the step up costs 18% to 42% more for roughly 24 to 30 extra square metres. In Punggol and Sengkang the “Executive” is a 126–134 sqm Premium Apartment rather than a maisonette, and the step is only about 3%, so there the Executive is often the better buy. Bear in mind HDB no longer builds Executive flats, so the stock only shrinks, and in August 2026 there were 170 Executive resales nationally against 578 five-room.

How many 5 room HDB flats are sold each month?

578 in August 2026, which was 22.9% of all 2,521 HDB resale transactions that month. The four-room is the biggest segment at 45.8%. The five-room is comfortably liquid in the large towns — Sengkang did 61 in the month, Tampines 56, Woodlands 51, Punggol 50 — but thin in the mature central towns, where Queenstown recorded seven and Ang Mo Kio nine.

Can I use my CPF to buy an older 5-room flat with a short lease?

Only partly, and this is the trap at the cheap end of the market. The remaining lease has to cover the youngest buyer to age 95 for CPF usage and loan tenure to work normally; below that, the amount of CPF you can use is pro-rated and the maximum loan tenure shortens. A 112 sqm five-room at $420,000 with about 49 years of lease left is excellent value for a cash buyer in their sixties and a financing problem for a buyer in their early thirties. Work out your own pro-ration before you view, not after you have made an offer.

How I got these numbers, and what they can’t tell you

Everything above comes from HDB’s own resale registration records, published as open data at data.gov.sg. I pulled every five-room and Executive transaction for July and August 2026 in twelve towns, town by town and month by month, and reconciled each cell two ways: against the registry’s own reported total for that filter, and against my own count of the transcribed rows. All twenty-four five-room cells agreed both ways. Price per square foot is my own arithmetic throughout — HDB has never published a psf figure for resale flats, so every HDB psf number you see anywhere, mine included, is a third-party derivation.

The limits, stated plainly:

• Two months is two months. The stability section is an argument for exactly this caution. Thin-town figures should be read as ranges.
• Twelve towns, not twenty-six. My sample is 379 of the 578 national five-room sales in August — 65.6%. The fourteen towns I did not pull include some small ones where the five-room market is tiny.
• September is still accruing. The all-time record table includes September 2026 sales and the month is not closed, so that table can change.
• The record list starts in 2017, which is the limit of the current registry resource. Given price levels, it is very unlikely any earlier five-room sale would enter the top 25, but I have not verified that and am flagging it rather than asserting it.
• The vintage control is genuinely inconclusive for this flat type, for the reasons in section five. I would rather say so than publish a clean-looking number built on three transactions.
• Registered prices are not asking prices. Everything here is what changed hands, which is the right basis — but it lags the market by a few weeks, and it says nothing about what is on the market today.