Home Prices 'Fell' in 537 US Counties. Nearly Half Got More Expensive.

If you own a house in Pueblo County, Colorado, you were told this year that home prices there fell. The median sale price went from $323,375 to $312,500, down 3.4%.

Over the same twelve months, housing in Pueblo County got 9.2% more expensive per square foot.

Both numbers are correct. They come from the same source, the same county and the same year. Only one of them is telling you anything about what your house is worth.

We checked every county Redfin reports. Pueblo is one of hundreds where the headline number and the underlying one point in opposite directions, and they are not exotic places.

Scott County in Iowa, around Davenport. Trumbull County in Ohio, around Warren. Linn County in Iowa, Jackson County in Michigan, Tippecanoe County in Indiana. Ordinary, affordable places where somebody reads the local paper and decides what their biggest asset is worth.

In Jackson County the median slipped 1.3% while the per-foot price climbed 9.3%. A homeowner reading only the first number would conclude their market went backwards in a year it got meaningfully more expensive.

A median is not an average and it is not a valuation. It is the middle sale. Line up every house that sold in a county this May, and the median is whichever one sits in the middle of that line.

That means it moves when the mix of houses changes, whether or not any house changed in value. A year with more starter homes and fewer four-bedrooms will show a lower median even if every single property in the county gained value.

Price per square foot is not immune to that, but it is far closer, because it adjusts for the most important thing that differs between houses: their size. When the two disagree, the median is usually the one telling you about the sales mix rather than the market.

In the year to May 2026, 537 counties posted a falling median sale price. In 254 of them the price per square foot rose over the same period. That is 47% of the declines.

It is not a quirk of this year either. We ran it back to 2013:

In every year of the last fourteen, somewhere between 39% and 80% of the counties that reported falling prices were places where housing got more expensive per square foot. The long-run figure is close to half.

The two years that are almost entirely green are 2021 and 2022 — and they are the shortest bars on the chart, because almost nothing in America fell in those years. The handful of counties that did show declines were nearly all mix effects, which is why their green share reaches 78% and 80%. It has never dropped below 39% in any year.

The fewer homes a county sells, the more its median jumps around, because each individual sale carries more weight in deciding where the middle lands.

Measured over 2013 to 2019 — stopping before the pandemic, which moved every market at once and would flatter the comparison — a county selling fewer than 50 homes a year saw its median swing about 22% annually. One selling 500 or more swung about 6%. In the smallest markets a double-digit "move" can happen with nothing at all going on underneath it.

The largest counties tick back up slightly, to 6.7%, and that band holds only 21 counties — few enough that one of them moving is visible in the average. The pattern is strong rather than perfect, and it would be poor form to smooth that away in a piece about not over-reading small numbers.

This is where it stops being a statistics quibble.

A seller who believes their market fell 5% prices accordingly, and in a county where the per-foot value actually rose, that is money handed to the buyer for no reason. On a $300,000 house, a 5% flinch is $15,000.

A buyer does the mirror image, crossing a market off the list because it looks like it is sliding, when the homes there are quietly getting more expensive. So does an investor screening counties on a single year of price change, which is exactly how a screen is usually built.

And a homeowner who reads that prices fell simply believes they are poorer than they are. Nothing in their life changed, and the number that told them so was measuring which of their neighbours sold this year.

None of this means a small market is unreadable. It means the median is the wrong tool for one, and better tools are sitting next to it in the same report.

Price per square foot is the first fix, and it is the one that catches the Pueblo problem, because it adjusts for the size of whatever happened to sell.

Sale-to-list is the steadier one. It is what homes sold for as a share of what sellers asked, worked out for each sale and then averaged, so one unusual house barely moves it. In the same thin counties where the median swings 21.8%, it swings 2.1%.

Read together they answer different questions. Sale-to-list tells you who currently has the upper hand in a negotiation. Price per square foot tells you what the housing itself is doing. The median tells you what sold.

Real declines exist, and plenty of the 537 are real: in slightly more than half of them the per-foot price fell too. Two measures agreeing is a much stronger signal than either alone.

Price per square foot is also not a perfect fix. It adjusts for size and nothing else, so a year that trades more condos, or more fixer-uppers, can still push it around.

And every figure here needs a county to have sold at least 20 homes in both years being compared, which leaves out the very thinnest markets in the country. The problem is worse than measured in exactly the places most likely to be misread.

Next time you see that home prices fell in a place you care about, ask two questions before you believe it.

First: what did the price per square foot do? If it went the other way, you are looking at a change in what sold, not a change in what things are worth.

Second: how many homes does this place actually sell? Under a hundred a year, one year of median price is close to a coin toss, and the honest answer is that you have not learned much either way.

If your county is one of them, you can look it up rather than take our word for it: here are all 537 counties that reported a falling median in the year to May 2026, with what the price per square foot did over the same year, how many homes each sold, and the two median prices being compared. 254 of the 537 got more expensive per square foot while their headline number fell.

P.S. The uncomfortable part of this for us is that we publish county medians too, on every market page. The number is not wrong, it is just answering a narrower question than people think it is, and that is on whoever presents it rather than on the reader.

Metrics, all per Redfin data at county level and sampled each May from 2013 to 2026: median sale price, median price per square foot, sale-to-list ratio and homes sold. A county qualifies for a year only when it sold at least 20 homes in that May and in the May a year earlier, since every figure here is a change between the two; the named example counties each recorded at least 150. The median and per-square-foot volatility figures use the 937 counties that qualified in every year from 2013 to 2019, which requires 2012 as well; sale-to-list is reported less often, so its figures rest on the 918 counties that qualified for it. Charts are generated from the production database and stamped with their as-of date. See the methodology.