Siem Reap Housing: Prices, Supply and Data Coverage

There is now enough public inventory in Siem Reap to move beyond a handful of anecdotal examples. On 29 September 2026, Realestate.com.kh showed 625 results in its broad Siem Reap sale feed, with prices displayed on 617 of them. That total is not a census of homes, however: the same feed includes 211 land listings, and the portal’s property-type categories are not mutually exclusive. The 625 figure is therefore useful as a measure of visible online inventory, not as a count of unique houses and apartments in the city.

The segment filters are more informative. On the same date, the portal showed 33 apartments or condos for sale, 126 villas for sale and 863 apartments or condos offered for long-term rent. Those pools can support asking-price and asking-rent benchmarks, but they do not reveal the prices at which transactions or leases actually close. Duplicate agency listings, project-level cards and unit-level cards can also make the visible inventory broader than a clean set of unique comparable properties.

Professional research provides a second, different layer of evidence. Knight Frank’s H1 2025 Cambodia review counted 3,849 landed-housing units across 27 Siem Reap projects and 589 condominium units across four projects. That is a tracked supply snapshot, not a live portal feed. Knight Frank’s later H2 2025 Cambodia report does not contain a dedicated Siem Reap residential chapter, so the H1 study remains the latest detailed city-specific supply benchmark from that firm that we found; it should not be presented as a 2026 register.

Official statistics form the third layer and also show where the evidence stops. The National Bank of Cambodia publishes its Residential Property Price Index for Cambodia, Phnom Penh and the rest of the provinces, but the published breakdown does not provide a separate Siem Reap city index. Population and visitor-arrival data help describe the city’s wider context, yet they do not substitute for closed-sale prices, achieved residential rents or vacancy data. The defensible conclusion today is therefore narrower: Siem Reap has enough visible inventory for segment-specific asking benchmarks, but not enough public transaction evidence to reduce the whole housing market to one city-wide price.

All figures below are asking-price or asking-rent indicators checked on 29 September 2026, not closed transaction prices. Listing counts show the size of the public sample, not a guaranteed count of unique properties.

The cleanest public sale benchmark comes from the apartment-and-condo filter rather than the city-wide feed. On 29 September 2026 it contained 33 listings, 32 of them priced. The portal displayed a $135,000 median asking price, a typical range of roughly $89,000 to $293,000 and a per-square-metre indicator of about $1,803. That is more useful than a few cherry-picked examples, but it is still a small market slice: project, age, condition, unit size and title status can move a condo well away from the median.

The villa sample is much larger, with 126 priced listings. The current category page shows a median of about $265,000 and a visible range of roughly $150,000 to $500,000. Per-square-metre comparisons are less straightforward for landed property because the denominator may refer to the building, the gross floor area or the land plot. Two villas with similar headline prices can therefore represent very different combinations of house size, land, condition and location.

Long-term apartment and condo rentals provide the deepest public sample: 863 listings. The median asking rent was $400 per month on the check date, with a typical range of about $283 to $650. A large listing pool makes that benchmark less fragile than a sale median based on a few dozen units, yet it remains an asking-rent indicator. The public feed does not show negotiated rents, vacancy periods, furnishing differences, agent fees, utilities or the share of the year for which a particular unit is actually occupied.

The city-wide $156,900 sale median is much less useful for a property-level decision. It sits on top of a broad feed that mixes condos, houses, villas and land, with property categories that can overlap. That figure can describe the centre of the visible portal distribution, but it should not be treated as the price of a “typical Siem Reap home”. A condo buyer needs condo comparables; a villa buyer needs landed comparables that are similar in both building and land characteristics.

Siem Reap’s product mix is one reason a single city price is a weak benchmark. Knight Frank’s H1 2025 professional snapshot counted 3,849 landed-housing units across 27 projects, compared with 589 condominium units across only four projects. The same report placed landed housing at roughly $430–$1,700 per square metre of gross floor area and condominiums at about $1,100–$2,700 per square metre. Those are different products measured within different supply structures, so combining them into one average price per square metre would obscure more than it reveals.

The difference is not merely numerical. A condo is relatively standardised for comparison: unit area, floor, layout, fit-out, building and title can all be matched. A villa or house bundles the building with a land component, making plot size and land rights part of the economic value. A $250,000 condo and a $250,000 villa can therefore sit in completely different competitive sets even before location and condition are considered.

The sources also define the condo market differently. Knight Frank tracked four condominium projects, while IPS wrote in 2025 that Siem Reap had three strata-titled condo projects. That gap does not, by itself, show that either source is wrong. Publication date, project phase, what counts as an active project and whether a completed strata title is required can all change the count. For a small segment, the definition matters because adding or removing a single project can materially alter the apparent market size.

For a real purchase decision, a small set of genuinely comparable properties is often more informative than a city-wide median. A condo comparison should stay close to the same projects, unit sizes, condition and title status. A villa comparison should match both house and land characteristics. The further a property sits from that peer group, the less meaning a broad headline number carries.

For a foreign buyer, the condo-versus-villa distinction is not only about price or lifestyle. Cambodia’s 2010 co-owned-building framework allows qualifying foreigners to own eligible private units in qualifying co-owned buildings, while the underlying land does not become foreign-owned. A condo with the right legal status and a villa sitting on land are therefore different assets before their prices are compared.

Budget alone cannot answer which one is the better fit. With a condo, the building status, the specific unit title, floor and foreign-ownership conditions need to match the legal framework. With a villa, the buyer needs to understand separately what right applies to the building and how the land interest is structured. The marketing label and the listing price do not resolve those questions.

For this market snapshot, the practical point is limited but important: price comparisons should start only after the legally available asset form is clear. The full purchase process and ownership structures belong in the dedicated foreign-buyer guide. On an individual property, the title and transaction documents matter more than the category name used in an online listing.

Tourism clearly matters to Siem Reap. Visitor flows affect employment, business revenue, seasonality and demand for parts of the accommodation market, so official arrival data is useful economic context. What it does not tell us is how many people sign six- or twelve-month residential leases, what rent they ultimately agree, or how long a particular apartment remains vacant.

The link is easy to overstate because several accommodation markets sit side by side. A visitor may stay in a hotel, guesthouse, serviced apartment or short-stay rental; a long-term tenant is making a different housing decision for different reasons. Even a strong recovery in visitor numbers does not automatically translate into higher long-term apartment rents or high occupancy for an ordinary residential unit.

Tourism statistics also say little about resale liquidity. A future buyer for a condo depends on the depth of that specific segment, the project, legal structure, price and available financing rather than the total number of people visiting Angkor. A defensible rental-return calculation needs achieved rent, real occupancy, operating costs and acquisition price. A defensible liquidity claim needs evidence from actual sales and time on market.

Tourism is therefore best used as background to the city’s economy, not as a housing-price formula. Without a measurable link between visitor growth and long-term residential outcomes, turning one into a forecast for the other would go beyond what the evidence supports.

The largest gap is the absence of a public, regularly updated series of closed residential transactions for Siem Reap itself. Portals show seller and landlord expectations, professional reports show the supply they track, and the National Bank’s published price index does not isolate the city. Together, those sources can produce a useful market picture, but they cannot establish a precise quarterly transaction median or prove that Siem Reap prices rose by a particular percentage.

A second limit is sample structure. Thirty-three condo and apartment sale listings are enough for an indicative benchmark, yet the median can move materially when only a few units enter or leave the market. Villa inventory is broader, but the properties vary much more in land and building characteristics, while portal categories can overlap. A comparison becomes stronger by narrowing the set to genuinely similar properties, not simply by adding more dissimilar listings.

Yield claims require even more care. Dividing the city-wide median rent by the city-wide median sale price does not produce a meaningful “Siem Reap yield” because the two medians can represent different properties and different segments. Even a published gross yield is not the owner’s net return: vacancy, management, repairs, taxes, commissions and negotiation at purchase or lease can all change the outcome. A property-level return needs a property-level purchase price and a realistic rental scenario.

Professional sources should not be forced into artificial agreement either. IPS’s three strata-titled condo projects and Knight Frank’s four tracked condominium projects may be measuring different sets. Until the definitions and cut-off dates align, collapsing those counts into one supposedly correct number would create false precision. The available evidence is strongest when used for scale, product mix and bounded asking-price ranges — while leaving transaction performance and individual-property economics to a narrower set of comparables and documents.