Average rent is not one measurement. At least three Canadian numbers travel under that name or something close to it, and each samples a different set of buildings, a different set of tenants and a different moment in time. Price a unit against the wrong one and it looks overpriced when it is fine, or fine when it is leaving money on the table.
The three numbers, and what each one counts
CMHC average rent
CMHC's Rental Market Survey reports a weighted average of all units in its universe, vacant and occupied together. That universe is privately initiated apartment and row structures with three or more units that have been on the market at least three months. Social and affordable housing sit outside it. The survey runs in the first two weeks of October, so the published figure is a snapshot of one fortnight, not an average of the year.
CMHC also splits the same stock several ways, and the splits are more useful than the headline. There is rent for occupied units, rent for vacant units, rent for units that turned over in the previous twelve months, and rent for units that did not. The published methodology defines each.
Asking rent
An asking rent is what someone is advertising today on a unit a renter can take. CMHC's vacant-unit series is an asking rent measured on the purpose-built stock it already surveys. Commercial listing trackers publish asking rents drawn from their own listing inventory, which is a different stock again and usually tilts toward condos and larger cities. Asking rent reacts fastest to market conditions, because every observation in it is a live offer rather than a lease signed years ago.
The CPI rent index
Statistics Canada's rent index is not a dollar level at all. It is a measure of price change for rented accommodation across the whole renting population: new tenants taking over a unit, tenants renewing a lease, and tenants sitting in the middle of one. Rent data come from a supplementary questionnaire on the Labour Force Survey, and roughly 8,000 renting households feed the monthly calculation. Since January 2019 the index has used a hedonic characteristics approach, so improvements to the quality of the stock do not read as pure price increases. StatCan sets out the method in its overview of shelter in the CPI.
Why the three can move in opposite directions
Turnover is the whole answer. In a province with rent control, a sitting tenant's rent moves by the guideline each year. A new tenant's rent resets to whatever the unit will fetch. So the average rent paid across a building keeps climbing on turnover even in a year when the rent being advertised on empty units is falling.
That is not a hypothetical. CMHC's 2026 Mid-Year Rental Market Update, published June 9, 2026, describes asking rents easing in most major markets while average rents on occupied units continued to rise, driven mainly by higher rents at turnover. Both statements are true at the same time because they are measurements of different populations.
StatCan makes the same point from the other direction: indicators built on advertised prices for vacant units can overstate the rental costs faced by renters as a group, because rent control limits what a renewal can rise by. Neither source is wrong. People quote them as if they answered the same question.
The stock question most people skip
CMHC's headline average covers purpose-built rental of three units or more. A condo rented out by its owner, a basement suite, a rented semi, a scattered portfolio of houses: none of that is in the number. CMHC surveys the condo and single-family rental stock separately, through the Secondary Rental Market Survey.
If you manage single-family or condo rentals and you benchmark your rents against the purpose-built figure for your city, you are comparing yourself to a stock you do not operate in. Check which survey a number came from before you let it change a price.
Comparing two years is its own trap
Subtracting last year's published average from this year's is not the same as CMHC's published percentage change. The percentage change is calculated on the same sample of structures, which removes the compositional effect of new buildings entering the survey, conversions, and sample rotation. A raw difference between two published levels carries all of that noise. New supply skews expensive, so a market that adds a lot of new buildings can show a rising average rent while every existing unit holds flat.
CMHC is direct that even the percentage change retains some compositional effect, from renovated units and from turnover. Treat it as a good estimate of rent movement rather than a clean reading of what an existing tenant experienced.
Which number belongs in which decision
- Setting the asking rent on a vacancy. Use asking rents for comparable units in the same stock, in the same submarket, right now. CMHC's vacant-unit and turnover series are the closest primary equivalents for purpose-built buildings.
- Budgeting next year's income on an occupied portfolio. Use your own rent roll and your province's guideline. If you need an external benchmark, the occupied or non-turnover series is the comparable one, not the asking rent.
- Estimating what a turnover is worth. Compare the turnover and non-turnover series for your market. The gap between them is the size of the reset you capture when a long-tenured unit finally comes back.
- Supporting a position with a lender, an owner or a tribunal. Use a primary source with published methodology, and name the series and reference period in the sentence. "CMHC average rent for two-bedroom purpose-built units, October 2025" survives a question. "Average rent in the city" does not.
Two questions before you use any rent number
Which stock does it cover? Purpose-built of three units or more, the secondary market of condos and houses, or one company's listing inventory. If you cannot answer, you cannot compare it to anything.
Which tenants, at which moment? Everyone in place, only the units that changed hands, or only the units sitting empty on the day of the count. An average across all tenancies tells you about a portfolio in motion. An asking rent tells you about today.
Most rent arguments inside a management company are two people quoting two different measurements at each other. Naming the series ends the argument faster than adding another number to it. The same discipline applies to the vacancy figures those rent numbers sit beside, which carry their own definition problem: see economic vacancy versus physical vacancy.