What 8,912 California Tax Auction Results Show About What Actually Sells

Every California county publishes a list of the tax-defaulted parcels it intends to auction. Almost none publish what happened afterwards. Most sales are held in person, results appear as a PDF on a county website for a while and then disappear, and no one aggregates them.

So we built the aggregate. This report covers 8,912 recorded parcel outcomes across 21 counties and 67 separate sales, from 2013 through the San Bernardino sale of 17 July 2026. Together those parcels changed hands for $98.9 million.

Everything below is counted from those records. Where the data does not support a claim, we say so instead of filling the gap — and there is a lot it does not support.

Four in ten parcels don't sell

The first useful fact is the plainest one. Of 8,912 parcels offered, 3,577 sold — 40.1%. The rest drew no bid at all.

That is the number to hold onto before any discussion of returns. The modal outcome at a California tax auction is not a bargain purchase. It is a parcel nobody wanted at the opening price.

One variable separates sell-through, and it isn't price

The opening bid at a tax sale is set by what is owed in back taxes and costs, not by what the land is worth. So the interesting quantity is the ratio between the opening bid and the county assessor's own valuation of the parcel.

We can compute that ratio for 5,065 parcels — those where the county published both an opening bid and an assessed value, excluding timeshare interests and agreement sales (see the method notes at the end). Sorted into bands:

Opening bid vs assessed value	Parcels	Sold	Sell-through
Under 0.25×	594	540	90.9%
0.25× – 0.5×	624	531	85.1%
0.5× – 1×	874	563	64.4%
1× – 2×	867	400	46.1%
Over 2×	2,106	397	18.9%

The gradient is clean and it runs in one direction across all five bands. A parcel opening below a quarter of its assessed value sells roughly nine times in ten. A parcel opening above twice its assessed value sells fewer than two times in ten.

This is also the single most common mistake we see in tax-sale write-ups: 2,106 of the 5,065 parcels — 41.6% — open above twice their assessed value. A great many tax-defaulted parcels are not cheap relative to what the county thinks they are worth. The unpaid taxes have simply accumulated past the value of the land.

The caveat that matters more than the finding

That table is mostly one county.

San Bernardino accounts for 4,486 of the 5,065 comparable parcels — 88.6%. Worse, its share is not evenly spread: it supplies 54% of the cheapest band but 96% of the most expensive one. So the two ends of the curve are not really being compared on equal footing.

Removing San Bernardino entirely leaves 579 parcels across the other seven counties:

Opening bid vs assessed value	Parcels	Sell-through
Under 0.25×	271	84.5%
0.25× – 0.5×	84	76.2%
0.5× – 1×	104	66.3%
1× – 2×	42	54.8%
Over 2×	78	41.0%

The direction survives. The magnitude does not. Outside San Bernardino the spread narrows from 91%-to-19% down to 85%-to-41%. The ratio still ranks parcels correctly, but it separates them far less sharply than the headline table implies.

Anyone quoting "90% of cheap tax-sale parcels sell" from data like this is quoting San Bernardino's high desert, where thousands of low-value lots turn over in bulk. Treat it as one market, not as California.

The county sets the level; the ratio only sets the rank

Here is the same data cut by county. Only eight of the 21 counties produce a single usable comparable at all.

County	Comparables	Overall sell-through	Under 0.5× sells	Over 2× sells
San Bernardino	4,486	44.9%	90.2% (n=863)	18.0% (n=2,028)
San Diego	231	95.2%	94.9% (n=176)	94.4% (n=18)
Placer	141	67.4%	82.7% (n=81)	13.3% (n=30)
Mariposa	96	24.0%	31.5% (n=54)	0.0% (n=6)
Sonoma	69	73.9%	97.0% (n=33)	50.0% (n=2)
Monterey	27	85.2%	100% (n=7)	72.7% (n=11)
San Francisco	14	28.6%	66.7% (n=3)	18.2% (n=11)
Shasta	1	—	—	—

Two things stand out.

The county effect swamps the band effect. A cheaply-opened parcel sells 31.5% of the time in Mariposa and 90.2% of the time in San Bernardino. Those are the same band. Being in the cheapest band in a cold county is worse than being in a middling band in a hot one — so a statewide base rate applied to a specific county can be wrong by a factor of three.

San Diego shows no gradient whatsoever. Cheap parcels sell 94.9%; expensive parcels sell 94.4%. Essentially everything clears. We flag this as a genuine counter-example rather than smoothing it away, with the honest qualifier that its expensive band holds only 18 parcels — too few to be certain of anything. San Diego's tax roll is also dominated by timeshare interests, which we exclude, so what remains is a small and unusual residue.

Why we will not predict what a parcel sells for

The obvious next question is what a given parcel will fetch. We tested it, and the answer is that the data does not support an estimate worth publishing.

Among 2,430 parcels that sold with both an opening bid and an assessed value, the sale price as a multiple of the opening bid:

Group	Parcels	25th pct	Median	75th pct	IQR spread
All sold comparables	2,430	1.01×	1.90×	3.81×	3.8×
San Bernardino	2,014	1.02×	1.81×	3.39×	3.3×
San Diego	220	1.00×	2.67×	5.58×	5.6×
Placer	95	1.00×	2.58×	5.26×	5.3×
Sonoma	51	1.20×	1.81×	4.00×	3.3×

The typical sold parcel goes for about 1.9× its opening bid. But the middle half of outcomes runs from roughly 1× to nearly 4× — a spread of 3.8×. A quarter of parcels sell at essentially the opening bid; a quarter sell for more than 3.8× it.

Narrowing the comparison does not help. Take the tightest slice available — one county, one sale date, one land-use class: San Bernardino's 4 June 2024 sale, 615 sold parcels, all classified "Land only." The interquartile spread is still 3.1× (1.03× to 3.19×).

So a point estimate would be wrong by a factor of three about a quarter of the time, even with 615 near-identical comparables. That is why nothing in this report, or in the analysis on our county pages, states what a parcel is worth. A base rate with its sample size attached is honest. A predicted price is not.

Timeshare interests are a different asset

Of the parcels we can match to an assessor record, 2,960 are timeshare interests — mostly fractional weeks in San Diego resorts. They sell at 23.9%, against 48.0% for everything else.

They are excluded from every figure above. Mixing a fractional resort week into a land base rate produces a number that describes neither.

What this data cannot tell you

• 13 of 21 counties produce zero comparables. They publish results but no assessed values that we can match. Shasta County's own GIS carries no assessor valuations at all; Trinity County's parcel layer is planning data only. For those counties we can report what sold, but not how the opening bid compared to value.
• Counties self-report, and some get it wrong. We have found published result sets with incorrect sale dates and prices, and county PDFs where a single record spans five parcel numbers.
• Date precision varies. Some counties publish results by fiscal year only, with no sale date. Those are recorded as year-precision rather than being given an invented date.
• Nothing here measures profit. Sale price is not resale value, and it says nothing about access, buildability, easements, back assessments, or whether the parcel is landlocked. Many low-priced lots are cheap for reasons that survive the auction.
• These are historical base rates, not forecasts. Bidder behaviour changes. A 2013 Placer sale is weak evidence about a 2026 one.

Method

Sources. County-published result lists (PDF and web), plus results from the auction platforms counties use, principally Bid4Assets and mytaxsale.com. Parcels are matched to county assessor and GIS records by APN to attach assessed values, acreage and land use.

Comparable set. A parcel enters the 5,065-parcel comparable set when the county published an opening bid above zero and an assessed value above zero, and the parcel is not a timeshare interest.

Agreement sales excluded. California's Chapter 8 process lets public agencies and certain parties buy a parcel by agreement, sometimes for a nominal sum — a $1 sale against a $2,900 opening bid. Five such records are excluded from price statistics, since they are real dispositions but not market prices.

Grouped lots excluded. Where a county offers several parcels as one lot with a single price, we record the lot and exclude it from per-parcel statistics rather than dividing the price across parcels.

Sell-through means the county recorded the parcel as sold. Assessed value is the assessor's total valuation, falling back to land value where no total is published.

---

This is a statistical summary of public records, published for research purposes. It is not investment advice, an appraisal, or a recommendation to buy or bid on any property. Base rates describe what happened to past parcels; they do not predict what will happen to any specific parcel. Do your own due diligence.