Houses vs Units in Australia: What Around 40,000 Sales Show

Start in Carlton, because it shows the unit problem better than any slogan.

A one bedroom apartment at 1209/6 Leicester Street sold in June 2011 for $428,000. On paper, it had the checklist buyers are told to like: rent from day one, walking distance to Melbourne Uni, inner city location, no obscure regional risk.

It sold again in April 2026 for $255,000.

The reflex reaction is "this is what apartments do". That reflex is wrong, or at least it's only half the story. The same database that shows that Carlton resale also shows a small block walk up in Norman Park, inner Brisbane, doubling over the same window, and an inner Adelaide villa selling 8% above ask in a slow market. "Unit" is doing all the work in the sentence, and it's hiding four different markets.

We pulled around 40,000 sold properties from the PropRadar database (sold price recorded, gross yield between 0% and 15% to drop micro apartment noise), matched roughly 27,000 of them to asking prices, then sliced them by what the listing says the property is: house, apartment, townhouse, villa, or the catch all "unit" label. Apartments split further by building density: small block, medium block, or tower of 50 or more units. That split matters because large apartment buildings account for much of the damage people associate with the whole category.

The vocabulary problem

The word "unit" gets used four different ways. A buyer says they're "looking at units". A listing portal labels something as a Unit because it has a strata title. A planner counts apartments and townhouses together as attached dwellings. And growth indices typically group everything that isn't a freestanding house under "units" so the sample is big enough to plot.

That convenience is hiding the actual distribution. In the sales table, the five mainstream types look like this:

How we calculated this

The sales table uses 39,896 sold properties from the PropRadar database. We kept records with a sold price and a gross yield between 0% and 15%, which removes most micro apartment and student studio outliers that inflate paper yields. Of those sales, 27,470 are also matched to an asking price. Apartment, Unit, Townhouse and Villa follow the property type shown on the listing, not a planner's classification. For apartment density, we estimate the size of the building from the unit numbers found at the same street address: small block means fewer than 20 units, medium block means 20 to 50, and large building means 50 or more. That density field is available for about 84% of apartment sales, 69% of townhouse sales, 93% of unit sales and 87% of villa sales. Capital growth uses suburb level price change measures for 8,207 house suburbs and 6,910 unit suburbs; that growth dataset groups apartments, townhouses and units together as "units".

The houses versus units headline is real. But that "Apartment" row is doing a lot of work in the average. Splitting it changes the picture entirely. We'll get to that.

Houses vs units: the headline

Across our suburb growth dataset, the national median house gained 30.9% over five years and 79.2% over ten. The national median unit gained 28.4% over five and 45.9% over ten. The five year view is close. Over 10 years, houses are roughly 33 percentage points ahead.

10 YEAR CAPITAL GROWTH (median, national)

Houses

+79.2%

Units

+45.9%

MEDIAN GROSS RENTAL YIELD

Houses

3.49%

Apartments

4.94%

Yield bars scaled to a 10% reference line; growth bars to 100%.

Across the states and territories, the picture is mixed. Growth over five years has houses ahead in five of eight jurisdictions, units ahead in three:

These are statewide medians across all suburbs in our growth dataset, including regional QLD, regional WA, and regional SA, so headline numbers sit higher than capital city hedonic indices that exclude regional stock and control for property mix. The direction of the house versus unit gap is broadly the same on both methods; the magnitudes differ.

Two patterns matter. First, the VIC and ACT inversions say as much about weak house growth in this cycle as they do about unit strength. Second, the gap over 10 years still favours houses nationally by a wide margin.

The structural reason is the same one investors have been told for decades: nobody can manufacture more land, but anyone with capital and a development approval can manufacture more units. Supply is the asymmetry. We see it most clearly when we split "apartment" by how much supply each building represents.

Apartments split three ways

Listing portals call almost everything "Apartment" if it has a strata title and isn't a townhouse. That covers a six pack walk up in Maroubra, a 40 unit 1990s block in Hawthorn, and a 600 unit Docklands tower. The buyer experience, the lender experience, and the resale experience are not the same.

We classify apartment buildings by unit count: small block (fewer than 20 units), medium block (20 to 50), large building (50 or more). The gap between sale price and asking price splits sharply along that line:

AVERAGE SALE vs ASKING PRICE

Small block

+1.66%

Medium block

+0.84%

Large building

-1.81%

Bars scaled to a +/-10% reference. Negative reads as discount to asking.

Two apartments in the same suburb, on the same price tier, sold the same week, but one is in an eight unit walk up and the other is in a 200 unit tower. On average, the walk up clears above the agent's price; the tower clears below it. Across thousands of sales, that's not coincidence, it's the structural difference between buying a slice of scarcity and buying one of 200 near identical neighbours.

This split is the useful part of the "apartments don't grow" debate. Blend strong small block stock with weak tower stock and you get a mediocre national average that doesn't describe either market well.

Large apartment buildings: where the damage sits

Cut the apartment market further by state and building size and large VIC apartment buildings stand out:

Large VIC apartment buildings are the warning sign. A median just under $510K, 6.31% gross yield, and 3% average discount to asking should not be read as a simple bargain signal. In this corner of the market, high yield often means rent has held up while the asset price has not. The gap between asking and selling also suggests some vendors are still pricing to an old market.

The worst 20 large building apartment postcodes in our sample (minimum 15 sales each) are dominated by inner Melbourne stock:

What it looks like at the listing level: a one bedroom CBD apartment selling for $145,000 in early 2026 (602/591-593 Elizabeth Street), a 16 sqm studio at 411/127-133 Leicester Street, Carlton selling for $169,000, and a 28 sqm flat at 213/408 La Trobe Street selling for $215,000. These are not automatic bargains. They're a class of stock with a small functional footprint, thin lender appetite, and a buyer pool that shrinks once people understand the building.

Three forces did most of the damage to this corner of the market. A building boom: large numbers of investor grade apartment projects were approved between roughly 2012 and 2017, most of which settled around 2016 to 2019. Prices stalled, then never properly re rated. A credit shift: APRA's 2014 investor lending cap and the 2017 interest only cap thinned the marginal buyer, and lenders became more cautious on new build apartments after the Banking Royal Commission. And a string of building defect events: Opal Tower, Mascot Towers, cladding inspections, insurance increases, and special levies changed how buyers and lenders looked at the whole asset class. NSW's Design and Building Practitioners Act 2020 helped restore some confidence, but trust takes longer to rebuild than regulation takes to pass.

The yield trap is the easy bit to spot. In our sample, 65% of studios and 92% of apartments under 50 sqm yield over 5%, but they also flag a suspicious_high_yield data quality warning at 10 times the rate of the typical apartment. Lenders are wary of units under 40 sqm. Many decline outright. That shrinks the future buyer pool, which puts downward pressure on price even when the rent is fine.

For a $500K apartment with a 7% headline gross yield, annual rent is about $35,000. A $7,000 annual strata bill on a tower with a lift, gym, and outstanding cladding work takes that to roughly $28,000 before council rates, water, insurance, vacancy, and management. The "gross" number is doing a lot of work.

Small and medium apartment blocks: different market

Small and medium apartment blocks do not look like Melbourne's CBD micro stock.

Apartments in small blocks (under 20 units) sold 1.66% above asking on average, with 48% clearing above ask, on a 4.58% yield. NSW small block apartments led at +3.15% above ask. QLD small block apartments hit +2.97%. WA small blocks: +9.66%. These are the walk ups, the 1960s and 1970s six pack blocks, the converted heritage stock. They don't have lifts, pools, concierge, or cladding obligations. The strata levies are small. The supply pipeline next door is constrained, because nobody is building a new six pack on an inner suburb block. They behave more like miniature houses than like the CBD towers everyone pictures when they hear "apartment".

The medium block segment (20 to 50 units) sits in the middle on both yield and discount. The market roughly tracks unit count: more units in the building means more substitutability, weaker pricing power, and a yield premium that compensates for weaker capital growth.

Inner Brisbane and inner Adelaide are the standout pockets where units have caught up with houses. The five year median in inner Brisbane (postcodes 4000 to 4179) was +95.8% for units vs +87.1% for houses, on a strong sample of 131 unit suburbs. Inner Adelaide (postcodes 5000 to 5099) was +54.2% vs +65.8%, narrower than the inner Melbourne and Sydney gap. Suburbs like Nundah, Clayfield, Stones Corner in Brisbane, and Park Holme and South Plympton in Adelaide, have seen unit prices roughly double over five years. Park Holme is at the upper bound of our growth signal, which only happens when a suburb has unambiguously doubled or more.

I would not project another doubling from here. That kind of move is a re rating, not a normal growth rate. But if you're looking at apartments in 2026, inner Brisbane and inner Adelaide are where the historical case is strongest. Inner Melbourne's CBD fringe and Sydney's large building investor belt are where it's weakest.

Townhouses: closer to a house than an apartment

Townhouses often get grouped with apartments because they're attached. The data points the other way:

A townhouse acts like a smaller, cheaper version of a house. The premium to asking is almost identical (+0.72% vs +0.91%), the share clearing above asking is identical to one decimal place (45%), and the yield sits closer to the house number than the apartment one. The structural reason: a townhouse owner usually has a courtyard, a roof they're responsible for, and a body corporate that covers a small handful of identical neighbours. Substitutability is much lower than in a tower; nobody is building another 200 of these next door.

By state, townhouses skew toward the eastern seaboard and ACT, with ACT showing the strongest yield (4.73%) and QLD the strongest gap above asking (+3.88%). NSW townhouses are the most expensive ($1.07M median, near house pricing) and the least likely to sell above ask (20%), consistent with a stretched buyer base.

The counterintuitive cut: when we split townhouses by complex density (small cluster vs large estate), the largest complexes (10 or more units) have the highest premium to asking (+1.70% average, 57% sold above ask). That's the opposite of the apartment pattern, and it's the supply story flipped. A large townhouse estate is usually a once off greenfield development, not a perpetual pipeline. Once it's full, nobody is building another 40 next door, because the land is gone.

Villas: the scarcity story

Villas are the smallest segment in our sample (636 sales, 256 matched to an asking price), so individual cells should be read with that in mind. But the pattern is consistent and worth calling out separately:

Villas sell 4.17% above asking on average, the highest premium of any mainstream type, and almost two thirds clear above ask. That's roughly 20 percentage points higher than the equivalent house number. They yield slightly more than houses (3.96% vs 3.49%) at a price about 18% lower. The trade off is that you don't get full land control, the complex usually has a small body corporate, and you can't make material structural changes the way you can to a freestanding house.

The reason for the premium is built in scarcity. Most of Australia's villa stock was built from the 1970s to the 1990s: single level, two or three bedrooms, on developable land in middle ring suburbs that planners would now class as a unit infill site. Approving a new villa complex in a built out suburb is hard. The supply pipeline is effectively flat. Combine that with a buyer pool that includes downsizers (who don't need a full house but want single level, single title living and outdoor space) and you get the strongest gap above asking in the dataset.

By state, WA villas led at +7.68% above ask, but the sample is small (88 sales). The safer read is that villas look consistently bid up where we have enough observations, and the national premium is much stronger than the house, townhouse, unit, or apartment premium.

How I would decide

Different filter logic for different types, in roughly this order.

For houses, the first filter is land or scarcity that can't be copied nearby. The growth thesis is built on the underlying land value, not the dwelling. If hundreds of similar houses could be added within a 20 minute drive over the next five years (think outer growth corridor with active subdivision approvals), the thesis is weaker even if the headline yield is decent.

For apartments in small blocks, the first filter is the building, not the suburb. Small block, ideally fewer than 20 units. No lift, gym, pool, or cladding remediation. Healthy capital works fund. Modest body corporate fees. If those boxes tick, this is the cohort where the historical numbers actually back the asset.

For apartments in large buildings, the maths is mostly defensive. I would treat this as a "no" unless the building is exceptional (limited competing supply nearby, no defect history, strong owners corporation), the finance is straightforward, and the buyer understands they're buying a yield asset rather than a growth one. The headline yields look good. The lender, strata, and resale pool problems can eat them.

For townhouses, treat them like a smaller house with strata costs. Read the body corporate disclosure properly. Check for big upcoming works. Confirm pet and short stay rules match your hold thesis. The gap above asking says the market treats well located townhouses with respect.

For villas, you're paying for scarcity. Expect to pay slightly above asking. The trade off is liquidity and rental demand, both of which the data supports. Don't expect house level capital growth; do expect to be able to sell when you need to.

The second filter, common to all of them, is holding power. Can you carry the cashflow on this specific property for five to seven years if rates stay uncomfortable? If yes, scarce stock has the better long term odds. If no, you're forced into the apartment market, in which case the small block cohort is where the historical case is strongest.

The third filter, new for 2026, is tax treatment. If the budget passes, established stock loses negative gearing for purchases after 7:30pm on 12 May 2026 and falls under the new CGT regime. New builds keep both. That's not a reason on its own to buy a new build, especially one in an oversupplied corridor, but it does mean a marginal cashflow case for an established inner ring apartment is harder to make than it was a month ago. Mechanics and worked examples in What the 2026 budget's negative gearing and CGT changes actually mean.

Frequently asked questions

Are units a bad investment in Australia?

It depends entirely which type. Apartments in small blocks and townhouses behave more like compact houses, sell above ask close to half the time, and historically gave up some growth in exchange for higher yield. Apartments in buildings with 50 or more units are where most of the damage in the headline numbers sits, especially in inner Melbourne, and the buying case is much narrower. Villas, the smallest segment, sell above ask more reliably than anything else in our sample.

What's the difference between an apartment, a unit and a townhouse?

Practically, an "apartment" is a strata flat in a residential building, usually multi level. A "unit" listing label in this dataset tends to describe smaller, lower density stock (a flat in a 1960s walk up, a 1970s converted house). A "townhouse" is multi level with its own entrance and usually some private outdoor space, typically in a small complex. Lender, body corporate and resale behaviour all differ, which is why the headline numbers split so cleanly.

What's a large apartment building?

In this analysis, it means any apartment building with 50 or more units. We estimate that from the unit numbers found at the same street address. The cutoff isn't arbitrary: the sale price versus asking price line breaks at around 20 and 50 units. The 50 plus cohort is where tower buildings, large strata costs and easy substitution start to dominate.

Do apartments grow as much as houses in Australia?

Nationally, no. Houses grew 79.2% over 10 years across 3,508 suburbs vs 45.9% for units across 2,514 suburbs. State data over five years is more mixed, with VIC, ACT and NT units edging houses recently. The "apartments don't grow" shortcut is too broad, but the underlying split matters more than the headline: apartments in small blocks and scarcity constrained suburbs have a much stronger record than large apartment buildings in oversupplied corridors.

Why do apartments yield more?

Rent doesn't scale linearly with price. In this sample, the median house sold for $1.06M at 3.49%, and the median apartment for $730K at 4.94%. The apartment costs about 30% less to buy, while the rent is often only modestly lower. That gives apartments more rent per dollar even when absolute rent is lower. The trap is when the yield looks great because the price has fallen, not because rent has risen.

Are townhouses a good investment?

The numbers are better than the apartment comparison suggests. Townhouse buyers paid slightly above asking on average and 45% of sales cleared above ask, both close to the house numbers and well above the apartment numbers. They yield about 1.5 times what houses do at a price roughly 18% lower. The trade offs are body corporate exposure (read the disclosure properly), a more constrained renovation envelope, and slightly lower long term capital growth than freestanding stock.

What about villas?

Villas are scarce and they sell like it. 64% clear above asking, the highest premium of any mainstream type, with an average of +4.17%. They yield slightly more than houses at a meaningful price discount. The catch is sample size, this is the smallest segment in our data, and the inability to expand or redevelop. If you can buy a well located villa, the resale liquidity is strong. They sell themselves, but you also pay for that.

What hurt Melbourne apartments so badly?

Oversupply, tighter investor lending, and building defect risk, all concentrated in larger apartment buildings. Melbourne approved a large volume of investor grade tower stock before 2017, especially around the CBD and inner north. APRA's investor lending restrictions thinned the buyer pool. The defect and cladding issues after 2018 hit confidence, insurance, finance, and resale liquidity. Those effects have lasted longer than many owners expected, and they show up most clearly in the large VIC apartment building numbers above.

Are apartments under 50 sqm worth buying?

Rarely. The headline yield can look excellent (median 6.01% in the 40 to 50 sqm band, 5.78% in the under 40 sqm band), but lender restrictions, thin resale demand and high strata costs usually wipe out the benefit. About 11% of studios in our sample flag a suspicious_high_yield data quality warning, vs less than 1% for typical apartments. The exception is a cash buyer who understands the building and doesn't need future resale liquidity, which isn't most investors.

Do strata levies really kill apartment yield?

They can. A small block with modest levies may be fine. A large tower with lifts, a pool, concierge, cladding works and a weak sinking fund is different. On a $500K apartment yielding 7% gross, annual rent is about $35,000. A $7,000 annual strata bill takes that to about $28,000 before council rates, water, insurance, vacancy and management. Always read the strata report before treating the gross yield as real.

How does the May 2026 budget change things?

If the legislation passes as announced, established residential property bought after 7:30pm on 12 May 2026 will lose negative gearing from 1 July 2027, and the 50% CGT discount will be replaced with cost base indexation plus a 30% minimum tax. New builds keep both. Existing owners and anyone under contract before the cut off are grandfathered. For someone looking at an established inner ring apartment with a marginal cashflow case, the deal is harder than it was. None of that addresses the supply, defect or strata issues that drove the underperformance in the data above, so the budget should change the cashflow maths, not the property selection discipline.

Keep reading: Where Australian property prices are rising and falling in 2026 for direction by state, Top 10 highest rental yield suburbs for the cashflow focused end of the market, Auction vs private sale for how sale method changes the buying maths, and Why rental yield matters for the underlying mechanics.

---

Analysis based on 39,896 sold properties from the PropRadar database, filtered to those with a recorded sold price and a gross yield between 0% and 15%. By type: 23,601 houses, 8,350 apartments, 3,770 units (the listing label), 3,720 townhouses, and 455 villas. Sale price versus asking statistics use the 27,470 of those sales that are also matched to an asking price. Capital growth uses the median across suburbs in our suburb statistics dataset: 8,207 house suburbs and 6,910 unit suburbs nationally, with sample sizes per state in the relevant tables. The suburb growth series groups apartments, townhouses and units together as "units". Apartment building size is estimated from the unit numbers found at the same street address; cutoffs are 20 and 50 units. State level growth figures include regional suburbs and sit higher than capital city hedonic indices. Inner city postcode aggregations cover roughly the CBD plus immediately surrounding suburbs: Melbourne 3000 to 3207, Sydney 2000 to 2100, Brisbane 4000 to 4179, Adelaide 5000 to 5099, Perth 6000 to 6079. All figures May 2026.

Budget references reflect the 12 May 2026 announcement and Treasury's published negative gearing and CGT explainer. The measures are subject to passage through Parliament; final legislation may differ from the announcement.

This is general information, not financial or tax advice. It does not take into account your objectives, situation, or needs. Property investment carries risk and individual outcomes can diverge significantly from medians. Talk to a licensed adviser before acting.