Auto Total Loss Frequency Hits a Record 23%, and the ACV Math Decides

A wrecked car does not need to be unrepairable to be totaled. It needs a repair estimate crossing a fixed share of its actual cash value, and in 2025 that happened more often than at any point on record: 23.1% of U.S. auto claims ended as total losses (CCC Intelligent Solutions, March 2026), up from roughly 20% in 2020.

That is a frequency shift wearing a severity number's clothes, and Progressive's June combined ratio of 90.0 is an early carrier-level reading of it.

Key Takeaways

• 23.1% of claims totaled in 2025, against roughly 20% in 2020. Each crossing swaps a negotiated repair estimate for a near-fixed ACV payout, changing the paid-loss distribution without any move in accident frequency.
• ADAS calibration appeared on 28.3% of repairable estimates at an average $485.56 per job, up from 21.8% a year earlier, which is why a bumper strike now generates an estimate a 2018 model year never would.
• Driveable total losses reached 10.4% of driveable claims, more than double five years earlier. Cars arriving under their own power are ending in the salvage yard.
• The Manheim index fell to 211.5 in mid-July 2026, down 0.6% from June, which raises total-loss frequency and cuts the salvage offset at the same time.
• Progressive's June combined ratio of 90.0 compares with 86.6, a 3.4-point deterioration, while policies in force grew 7% to 40.1 million.

Two Decision Rules, Not One National Parameter

Roughly 30 states declare a total loss once repair costs reach a fixed percentage of actual cash value, typically between 70% and 75%, with Florida at 80% and Colorado and Texas requiring repair costs to meet or exceed full ACV (WalletHub, 2026). Roughly 20 more, including California, Georgia, Illinois, Ohio, Pennsylvania and Washington, use a total loss formula comparing repair cost plus projected salvage against ACV.

The distinction is larger than it sounds. A vehicle worth $8,000 with a $5,000 repair estimate and $3,500 salvage value is repairable in a 75% threshold state, since $5,000 is 62.5% of ACV. The same claim is a mandatory total loss under a formula state, because $5,000 plus $3,500 is $8,500.

So the same collision on the same vehicle resolves differently by state, and salvage recovery expectations do underwriting work in roughly 40% of the country before an adjuster opens an estimate. A countrywide physical damage severity curve is a state-weighted mixture of two decision rules, and a book concentrated in formula states carries a structurally higher total-loss share than the national figure implies.

The Mix Shift Reads as Cost Inflation

Both inputs to the threshold track wholesale used-vehicle prices, and both have softened. The Manheim Used Vehicle Value Index fell to 211.5 in the first half of July 2026, down 0.6% from June on a mix, mileage and seasonally adjusted basis (Cox Automotive).

That produces an asymmetry rather than an offset. A falling index compresses ACV, which pulls more claims across a fixed repair-cost threshold. It also lowers what a totaled vehicle fetches at auction, trimming the salvage credit on every one of those additional total losses. Both effects push net paid severity upward, which is the opposite of what a reserve model treating salvage as a flat percentage of ACV would assume.

Fleet composition pushes from the other side. There are 12 million fewer vehicles six years old or newer in operation than in 2020, a legacy of the 2021 to 2023 production shortfall. Average total cost of repair rose only 1.7% in 2025 to $4,818, the smallest annual increase since 2017, but that blend conceals a split: ADAS calibration averaged $485.56 per job and appeared on 28.3% of repairable estimates, up from 21.8%.

A newer vehicle carries a higher ACV cushion and still crosses the threshold on what used to be a fender-bender, because a bumper strike that knocks a forward camera out of calibration generates an estimate a 2018 model year without those sensors never would. Driveable total losses hit 10.4% of driveable claims, more than double five years earlier.

Dimension	Repairable Claim	Total-Loss Claim
Primary cost input	Parts, labor, calibration fees	Actual cash value less salvage
Typical settlement speed	Longer, multi-step repair and supplement cycle	Faster, single valuation event
Dispute risk	Supplement and parts-sourcing disputes	ACV and salvage-value valuation disputes
Sensitivity to used-vehicle values	Low	High, on both ACV and salvage sides

Here is where the reserving arithmetic diverges from the headline. A rising total-loss share is claim-disposition frequency, but it surfaces in the data as severity trend, because a lump-sum ACV payout and a metered repair estimate are structurally different distributions being blended into one average. As total losses take more of that blend, mean paid severity rises even if the cost of an individual repair does not move at all.

CCC's bodily injury series shows the discipline this calls for: paid BI severity rose 10.3% year over year and 32% cumulatively over four years, a trend actuaries routinely decompose into litigation cost, medical trend and claim mix rather than treat as one homogeneous rate. Taking a blended physical damage trend at face value attributes a frequency-driven mix shift to cost inflation, and sizes the wrong lever in the rate indication.

Faster to Close, Harder to Value

The reserving consequence runs in two directions at once. Total-loss claims settle faster, with no repair cycle, no supplement process and no parts-and-labor sequence. Once ACV is agreed, the claim closes. As the total-loss share rises, average paid development for physical damage shortens mechanically, which can read as favorable development to anyone not tracking the composition underneath it.

Variance moves the other way. ACV determinations are contested more often than repair estimates, particularly in formula states where the salvage component is itself an estimate open to dispute, and state-level diminished-value and total-loss valuation litigation adds tail volatility to a line long treated as fully credible at short maturities. A triangle that develops to ultimate inside 90 days can carry meaningfully more per-claim variance even as its time-to-close compresses, because more of its claim count is a valuation negotiation rather than a repair invoice.

Deductible composition pulls the same way for reasons unrelated to repair technology. CCC's Kyle Krumlauf noted that "insurance coverage and claim filing behaviors have dramatically shifted as consumers react to an uncertain economic landscape." A policyholder with a $1,000 deductible on a vehicle worth $6,000 has less reason to contest a marginal total-loss determination than one carrying $250, since the net difference between repaired and totaled narrows as the deductible rises against ACV.

The rate side has not caught up. Industry auto premiums fell roughly 6% in 2025 as carriers passed through the 2023 and 2024 increases, with Insurify projecting only 1% for 2026 (Repairer Driven News), a trajectory that assumes the favorable severity environment behind the 2025 relief holds.

Progressive's second quarter ran 87.3 against 86.2, with policies in force at 40.1 million, direct personal auto up 10% and agency up 8% (Progressive, July 2026). Growing personal auto count at that pace while the monthly combined ratio worsens more than three points is what it looks like when earned premium has not caught a loss cost trend that moved after the in-force rates were filed.