Updated 2026-08-25 · data through July 2026

New vs. used car prices: 73 years in one chart

Used-car prices are up 28.7% since December 2019. So is everything else, to within 0.02 of a point — the pandemic premium is gone. New vehicles never joined the spike and never joined the fall.

The headline, as of July 2026

• +28.7%used cars since Dec 2019
• +28.7%everything else, same span
• +21.4%new vehicles, same span
• +41%used cars’ real peak, January 2022

Read those first two numbers together, because they are the story. Since December 2019 the used-vehicle index has risen 28.7% and the price of everything else has risen 28.7%. Strip out general inflation and used cars now sit level with where they were before the pandemic — 0.02 of a percentage point from it. At the peak in January 2022 they were 41% above it in real terms. That entire premium is gone.

New vehicles did not follow them up and have not followed them down. They are 21.4% higher than in December 2019, which against 28.7% inflation leaves them 5.7% cheaper in real terms than before the pandemic. Two vehicle markets, one shock, opposite paths through it.

Seventy-three years, three lines

In 1953 a car cost more, relative to the rest of American life, than it does now: the new-vehicle index opens the series at 46.3 against an all-items index of 26.6. The three lines run together through the inflation of the 1970s and early 1980s and then separate for good. Today the cost of living stands at 333 on that same base while new vehicles sit at 179 and used at 180. Since the mid-1980s, measured this way, cars have been one of the slowest-rising things a household buys.

That sentence carries a large caveat and it is worth reading before the chart gets quoted anywhere: these indexes hold quality constant. They are not the price on the window sticker. The section below spells out exactly what BLS does and does not adjust.

New-car prices ratchet. Used-car prices swing.

The two lines have different personalities, and it shows up cleanly if you count how often each index has fallen. Across 883 months, the used index has spent 14 separate stretches at least 5% below a previous high. The new-vehicle index has managed 4. Its deepest fall in seventy-three years is 12.9%, bottoming out in 1954. The used index has been 30.3% below a high — more than twice as deep.

The most striking stretch belongs to new vehicles: from October 1973 to August 1997 — 287 consecutive months, nearly 24 years — the new-vehicle index never once printed a reading below the same month a year earlier. The best run the used index has ever put together is 60 months. New-car prices behave like a ratchet. Used-car prices behave like a market.

The pandemic put that difference under a microscope. From February 2020 to January 2022 the used index rose 53% while new vehicles rose 13%. Then it reversed: from January 2022 to August 2024 used prices fell 19% — and new-vehicle prices, over the very same months, went up 5%. Nothing about the shortage that inflated used values ever translated into a discount on a new car afterwards.

What these indexes actually measure

Five things about the data, each from BLS’s own fact sheets and notices, because each one changes how the chart should be read. This is the section to check before quoting the chart anywhere.

The used index is a 2-to-7-year-old car. BLS states the item definition plainly: the used cars and trucks index “is comprised of used vehicles that are between two and seven years of age.” If the car you are selling is fifteen years old, this index describes the market above yours, not yours. Older vehicles track it loosely and lean far harder on condition, mileage and, at the bottom, scrap value.

Neither index is a sticker price, and neither is an average price. The new-vehicle index is built from a dealer transaction dataset that, in BLS’s description, “typically contains about 250,000 transactions per month.” The used index is not collected by BLS at all — “all price information for used cars and trucks used in the CPI comes from J.D. Power Valuation Services,” adjusted for depreciation each month. Both measure change, not level. BLS publishes no average price for a car in either category.

Quality is held constant, and that is the big one. When a model year adds equipment, BLS adds the manufacturer’s stated cost of that equipment to the old model’s price before comparing the two, so the improvement does not read as inflation. Its own worked example takes a 2020 model going to 2021 with three additions — smartphone integration at $50, a nine-speed transmission at $399, keyless entry with auto start/stop at $459 — and adjusts the 2020 price upward by those amounts, plus a markup factor, before measuring price change. Over seventy-three years of seat belts, airbags, anti-lock brakes, emissions equipment and backup cameras, those adjustments compound. It is why the index can show new vehicles rising more slowly than the cost of living while the sum a buyer actually hands over has climbed much faster. Both are true. The index answers “what would the same car cost?” — not “what does a car cost?”

This is cost-based adjustment, not hedonic. Worth stating because the two get conflated constantly, including in write-ups of this exact chart. Hedonic regression is a CPI technique, but not one used on vehicles. BLS draws the line itself: “The important difference between this cost-based quality adjustment method versus a hedonic quality adjustment method is the use of actual cost data provided directly from manufacturers, which are not estimated costs derived from a regression model.” And the two lines are not adjusted independently — a used vehicle carries forward the same dollar adjustments made to it when new, depreciated.

The series have seams, and one of them affects the volatility count above. Three dated changes matter here. With January 2018 data the used index “changed from a three-month moving average to a single month price change” — so before 2018 the used line is smoothed and its swings are understated relative to everything after. With April 2022 data BLS replaced its own new-vehicle price collection with J.D. Power transaction data and dropped motorcycles from the index. And there is no October 2025 reading for the cost-of-living line at all: BLS “did not collect CPI data from October 1, 2025, through November 12, 2025,” and marks that month with footnote X, its own code for “Data unavailable due to the 2025 lapse in appropriations.” Every comparison against general inflation on this page skips that month rather than filling it in. Both vehicle series were published for October 2025 and are unaffected.

The direction of that first seam is worth being explicit about, because it cuts against the finding rather than for it. Smoothing suppressed used-car volatility for the first sixty-five years of the series. The gap between 14 downturns and 4 is therefore a floor on the difference between these two markets, not an exaggeration of it.

What it means if you are the one selling

Three practical readings, in the order they usually matter.

The window where a used car outran inflation has closed. For roughly four years, a car bought before 2020 was worth more against its purchase price than any depreciation table said it should be. In nominal dollars some of that is still visible. In real terms it is finished: 0.02 of a point from December 2019. If you have been waiting for values to come back, the data says the wait ended rather than that it continues.

Falling used prices have not been a crash. The used index is 1.9% lower over the past twelve months and still 29% above December 2019 in plain dollars. That is a market returning to normal, not one collapsing, and the history above shows it has been 30% below a high before now.

Selling and replacing are two different trades. The chart’s most useful line for a seller is the one that barely moves. Whatever your car fetches, the replacement is coming out of a market that has fallen 5% from a high 4 times in seventy-three years and rose through the entire used-price collapse of 2022–2024. Waiting for new vehicles to get cheaper is a bet the series has almost never paid.

Methodology

Three published series, drawn as published, fetched from the BLS flat files on 2026-08-25: CUSR0000SETA01 (new vehicles), CUSR0000SETA02 (used cars and trucks) and CUSR0000SA0 (all items), each seasonally adjusted, U.S. city average, monthly from January 1953 to July 2026. Every percentage on this page is a ratio of two published index values; nothing is modelled and no month is interpolated.

Seasonally adjusted series are used because the headline comparison runs from a December to a July, and unadjusted indexes would carry the difference between those two months in the used-car sales year into the answer. The trade-off is worth naming: BLS recalculates seasonal factors each February and revises the previous five years, so figures here covering 2021 onward can shift slightly at the next revision. That is why the page carries a data-as-of date, and why the underlying series are named — anyone can pull them and re-run the arithmetic against a later vintage.

Sources

• BLS Consumer Price Index — series CUSR0000SETA01 (new vehicles), CUSR0000SETA02 (used cars and trucks), CUSR0000SA0 (all items), seasonally adjusted, U.S. city average
• Measuring Price Change in the CPI: New vehicles — item definition, transaction dataset and quality-adjustment method
• Measuring Price Change in the CPI: Used cars and trucks — the two-to-seven-year age definition and the J.D. Power source
• Quality Adjustment in the CPI: New, Used, and Leased Motor Vehicles — the cost-based method, the worked model-year example, and the stated contrast with hedonic adjustment
• CPI methodology change notices — used cars and trucks, posted 21 November 2017; new vehicles source change, posted 13 January 2022
• 2025 federal government shutdown impact on the CPI — the collection dates behind the missing October 2025 index
• BLS public data files — cu.data.1.AllItems and cu.data.14.USTransportation, fetched 2026-08-25; footnote code X defined in cu.footnote

Related: the used-vehicle index on its own, back to 1953 · which month of the year sells best · the scrap index that sets the floor under an old car.