Sihanoukville Housing: Prices, Supply and Data Coverage

Useful Sihanoukville housing data exists, but the main sources measure different things. Knight Frank’s H1 2025 research counted 11,742 existing condominium units across 21 projects: 6,377 high-end units, 3,366 mid-tier and 1,999 core. The same report recorded 1,757 landed-housing units. Those figures are a useful professional baseline for built stock under Knight Frank’s methodology; they are not a count of homes actively for sale today.

The live portal answers a different question. On 29 September 2026, Realestate.com.kh displayed 69 apartments and condos under its Sihanoukville filter, with a headline median asking price of $83,650, a stated typical range of $59,550–$194,500 and a median of $1,764 per square metre. The results themselves show why that headline needs care: Koh Rong, Prey Nob and Ream can appear under the broader location label, project pages sit beside individual units, and listing ages run from days or weeks to several years. The portal is therefore valuable as a source of individual asking offers, not as a ready-made transaction series.

Once geography and development stage are separated, several usable layers emerge. In the city core there are fresh ready-to-move units with disclosed price and floor area, although the number of close comparables is small. New and off-plan supply provides more current asking points, but much of the fresh evidence is concentrated in individual developments, particularly Time Square 10. Ream and Koh Rong form another visible cluster outside the city; their product types and price levels are different enough that folding them into a city median changes what the number represents.

Official series add context without solving the city-price problem. The National Bank of Cambodia’s Residential Property Price Index is published for broader geographies, including Phnom Penh and Rest of Provinces, rather than as a Sihanoukville condo index. The National Institute of Statistics’ 2019 census remains useful for dated administrative and demographic context, but it is not a current measure of housing demand. The government’s 2024 Sihanoukville programme confirms that stalled and unfinished projects are a distinct policy issue, yet those buildings do not automatically become active homes for sale.

The defensible result is therefore a set of separate measurements rather than a single market number: professional existing-stock estimates for their stated period, current asking evidence inside narrow comparable segments, and the visible composition of listings on the snapshot date. The same evidence does not provide a complete record of completed transactions. Keeping those layers apart is more informative than presenting one precise-looking average for the whole city.

The cleanest current example of completed city housing is a very small set of studios explicitly marked ready to move in. Two recent Sihanoukville city listings each ask $50,000, with gross floor areas of 47 and 49 square metres. That works out to roughly $1,064 and $1,020 per square metre, with a two-observation median of about $1,042 per square metre. The sample is far too thin to call this a Sihanoukville market price; it only shows where two directly comparable ready studios are currently offered.

A nearby result illustrates how quickly comparability can break. A Studio Residence in LZ SEA VIEW PREMIUM is offered at $54,715 for 45 gross square metres, or $1,216 per square metre, but the portal presents it as a new-development offer rather than the same kind of completed resale unit. Adding it to the two ready studios would make the dataset larger while making the category less coherent. A condo filter alone does not establish that the homes have the same sales stage.

Fresh evidence is more plentiful among new and off-plan city condos, but it is highly concentrated. Six current small-unit offers in Time Square 10 with disclosed total prices run from about $43,760 to $67,000, with a median of $52,461. Because every observation comes from one development, the result is project-dominated: it is useful for describing the asking level visible in Time Square 10 on the snapshot date, not for describing all new Sihanoukville condos. A per-square-metre median is not shown for this set because the unit cards do not establish a sufficiently consistent floor-area basis for that calculation.

Geography produces an even larger distortion. The broad portal filter also surfaces recent Platinum Coast units in Ream and Port Yves Palace homes on Koh Rong. Seven visible individual offers across those two clusters range from roughly $75,162 to $454,944, but a combined median would be misleading: they are different locations and product types, and four of the seven offers come from one project. Their value here is to show how province-level search results can reshape a city headline, not to enlarge the city sample with resort and island stock.

IPS Cambodia’s 2025 market article is useful as a secondary sense-check on price bands. It quoted roughly $1,400–2,500 per net square metre for high-end condos, $1,000–1,400 for mid-range and $700–1,000 for lower-priced stock. These are commercial asking-price bands, not a registry of completed sales. They can help place a listing within a published segment range, but they should not be treated as an independent transaction benchmark.

Price per square metre has its own data problem. Portal cards can refer to gross floor area, net internal area or land area, and sometimes the basis is not disclosed at all. Dividing the same asking price by different area definitions creates different per-square-metre figures without any change in the property itself. A calculated median is therefore meaningful only inside a group that shares the same area basis; missing area definitions are a reason to leave the median blank, not to estimate it.

The portal headline of $83,650 and $1,764 per square metre remains a useful description of what the broad Realestate.com.kh results page displays on 29 September 2026. After separating islands, Ream, project pages, older listings and different sales stages, however, the headline is clearly describing a mixture, not a homogeneous city segment. For a buyer comparing actual homes, a smaller transparent set is more useful than a larger number whose components are not comparable.

Snapshot as of 29 September 2026. Calculations are shown only where geography, stage and floor-area basis are sufficiently clear; a missing median means the records are not comparable enough, not that no properties exist.

A large portal count is not the same thing as the number of homes a buyer can genuinely compare today. The September 2026 results mix recently listed or updated properties with cards more than a year old and others that have remained visible for two, three or more years. Some of the freshest entries are development pages, not individual homes. An old card therefore cannot be assumed to remain available, while a new project page cannot be counted as one unique apartment.

The useful current layer starts with records that have a recent timestamp, a clear location and a specific property. Listings older than 12 months are better treated as legacy evidence: they show what projects and asking levels have appeared on the portal, but they should not silently shape a current median. Where no update date is disclosed, freshness is unknown, not presumed. This matters in Sihanoukville because deeper result pages contain a substantial number of listings that are two, three or four years old.

Duplication is not limited to an exact copy of the same advertisement. One unit can be marketed by several agents, a project page can sit next to a unit from that same development, and a standard floor plan may genuinely exist in many separate apartments. Removing every repeated price-and-size combination would therefore also be wrong. The user-facing consequence is simpler: the count of 69 describes cards in the portal interface; it does not prove that 69 unique city homes are simultaneously available.

Professional existing stock is a different base again. Knight Frank estimated 11,742 existing condo units at H1 2025. Those homes may be occupied, held off market, owned by many different parties or absent from the major portal entirely, so comparing 11,742 with 69 as if both were inventory counts would be meaningless. One figure describes accumulated stock under a research methodology; the other is one website’s current search output.

Future supply must also stay separate. Knight Frank wrote that condominium stock could rise to 31,824 units by 2029 if current projects resume and are realised; the wording is explicitly conditional. The consultant recorded one new H1 2025 condo launch, Time Square 10 Ocean View, with 1,000 units and expected completion in 2029. Current portal offers show the project being marketed while under construction, but future units are not part of today’s completed stock.

The same distinction applies to stalled and restarting buildings. Sihanoukville’s special government programme includes measures aimed at unfinished and problematic projects, confirming that this is a material category for the city. Participation in a programme or an announced restart, however, does not establish that a particular apartment is completed, operating and available for a current transaction. For live supply, the status of the specific property matters more than the size of the potential pipeline.

The largest gap is completed transaction pricing at city level for comparable housing types. Portals record the seller’s asking position, not the amount ultimately exchanged in a closed sale. Without a registry or a consistent professional transaction series, there is no defensible way to calculate a city-wide ‘discount to market’, average negotiation gap or typical completed-sale price per square metre. A fresh listing remains an asking price until there is evidence of the completed transaction.

There is also no single current city-level condo index for Sihanoukville in the official series used here. The National Bank of Cambodia’s index is useful for broader price direction, but its Rest of Provinces geography is wider than one city and its condominium stock. Relabelling that series as a Sihanoukville index would change what the statistic actually measures. A dated census cannot substitute for a current demand indicator either.

Public sources also do not provide a robust current city-wide measure of occupancy, vacancy, net rental yield, time to sell or liquidity. A developer may advertise a return and an individual building may report its own operating performance, but those claims belong to that product and that methodology. They cannot be averaged into a Sihanoukville market yield without an independent comparable dataset.

Nor does a portal provide complete active inventory. Some sales are marketed outside large platforms, the same home can appear more than once, and a card can remain online after its status changes. The cleaned n used on this page is therefore coverage of a particular public source on a particular date, not an estimate of every home that could theoretically be bought in the city.

None of these gaps is a verdict on whether the market itself is good or bad. They simply mark the boundary between evidence and inference. For a particular apartment, a small group of genuinely comparable recent offers is more useful than a precise-looking city average assembled from incompatible records.