CAFE Standards Explained: U.S. Fuel Economy Rules, 1975–2026

What CAFE is

Congress created CAFE in the Energy Policy and Conservation Act of 1975, after the Arab oil embargo. NHTSA sets the standards; EPA runs the fuel economy tests. Each automaker's required average depends on the mix of vehicles it sells: since the late 2000s, every vehicle has a target based on its footprint (roughly, wheelbase times track width), and the company must meet the sales-weighted average of those targets. Cars and light trucks have separate curves, and automakers can bank, trade and transfer credits to cover shortfalls.

So a headline number like the 2024 rule's 50.4 mpg for model year 2031 is NHTSA's projection of the industry-wide average on its lab test if automakers sell the mix it expects. It is not a minimum every model must meet. This page covers cars and light trucks; medium- and heavy-duty trucks have separate programs. (49 U.S.C. 32902)

Why CAFE mpg is higher than your window sticker

Three different mpg numbers get quoted, and they are not interchangeable:

• Lab compliance test (CAFE): EPA's original two-cycle test, weighted 55% city and 45% highway, plus any regulatory credits. Every CAFE standard on this page uses it.
• Window sticker: a five-cycle test that adds high speed, air conditioning and cold weather, meant to reflect what drivers will see.
• EPA real-world estimate: what EPA's annual Automotive Trends report uses to compare model years (five-cycle results, weighted 43% city and 57% highway).

NHTSA says real-world fuel economy is generally 20 to 30 percent lower than compliance values (90 FR 56438, footnote 17). EPA's example of a 2024 Toyota Prius shows the gap:

• 80 mpg (lab compliance test)
• 57 mpg (window-sticker rating)
• 56 mpg (EPA real-world estimate)

Source: EPA Automotive Trends Report 2025, Appendix C.

How the standards changed, 1975–2026

1975–1985: Congress wrote the first passenger-car schedule into law, rising to 27.5 mpg for 1985 cars on the lab test, and left truck standards to NHTSA. 1985–1989: the Reagan administration lowered the car standard for four model years. 1996–2001: appropriations riders barred NHTSA from changing the standards, freezing them. 2003–2007: the Bush administration raised light-truck standards and moved trucks to footprint-based targets, and President Bush signed the Energy Independence and Security Act (EISA), which required a combined average of at least 35 mpg by 2020 on the lab test. 2009–2012: the Obama administration finalized the 2011 standards (proposed under Bush) and then issued two joint rules with EPA covering 2012–2025. 2017–2020: the first Trump administration reopened the 2022–2025 standards and the SAFE rule slowed annual increases to 1.5%. 2021–2024: the Biden administration issued two rules raising them faster. 2025–2026: the second Trump administration reinterpreted the statute, Congress set the penalty to $0.00, and NHTSA proposed and then announced a reset.

What new vehicles actually achieved

EPA's real-world estimate for each year's new vehicles climbed from 13.1 mpg in 1975 to 22 mpg in 1987, slid to 19.3 mpg by 2004 as buyers shifted to SUVs and pickups, then rose to a record 27.2 mpg in 2024. The model year 2025 figure is preliminary.

Source: EPA Automotive Trends Report 2025, Table 2.1. These are averages for each year's new vehicles, not for all cars on the road. Before 1986, EPA adjusts the older lab results to estimate real-world mpg.

The vehicle mix changed as much as the vehicles. In 1975, sedans and wagons were 80.6% of new-vehicle production; in 2024 they were 23.7%, and SUVs built on truck platforms were 49.6%. Because light trucks have lower targets, that shift lowers the industry-wide requirement even when each curve stays the same.

Source: EPA Automotive Trends Report 2025, Table 3.2.

On the compliance test, the industry as a whole beat its required average in 12 of the 18 model years from 2004 to 2021 and fell short in 6 (2016, 2017, 2018, 2019, 2020, 2021). An industry-wide average does not show whether a standard constrained any one automaker, or whether a company that fell short covered the gap with credits.

Source: NHTSA CAFE Public Information Center, fleet performance (July 24, 2025 edition; final data end at model year 2021). From 2017, achieved values include air-conditioning and off-cycle credits.

New-vehicle prices and the 2008 break

The chart divides BLS's consumer price index for new vehicles by its index for all consumer prices. When the line falls, new vehicles got cheaper relative to everything else; when it is flat, they rose at the rate of inflation. From 1967 to 2008 the line fell by 58%. After its 2008 low it barely moved for several years, and it has fallen much more slowly since.

Our calculation from BLS series CUUR0000SETA01 (new vehicles) and CUUR0000SA0 (all items), not seasonally adjusted, annual averages over matched months, 2009 = 100. BLS data through August 2026; the current year is excluded until it is complete. *2025 averages eleven months: BLS never published October 2025 data because of the federal funding lapse.

Measured as a trend (a least-squares fit to the log of the index), the relative price of new vehicles fell about 3.3% a year from 1998 to 2008, fell about 0.4% a year from 2009 to 2016, fell about 0.7% a year from 2016 to 2025. These rates describe the data; they are not estimates of any policy's effect.

What this index measures

BLS tracks the price of the same vehicle models over time. When a model changes, BLS asks the manufacturer what the change cost. It treats some of those costs as quality improvements rather than price increases, and removes them from the index. That category includes fuel-economy improvements and federally required safety equipment.

Changes made only to meet air-pollution standards are handled the opposite way. For models introduced in January 1999 or later, BLS counts their cost as a price increase, not a quality improvement. Before 1999, BLS removed those costs too. By BLS's own estimate, the new-car index rose 174.2% from December 1968 to December 1997 as published; had pollution-control costs been counted as price, it would have risen 230.2%.

So this index shows price change after adjusting for quality. On its own, it cannot tell how much any regulation changed what buyers paid.

Sources: BLS, quality adjustment for vehicles; BLS, treatment of mandated pollution-control measures. BLS does not say how it classifies technology added to meet EPA's greenhouse-gas standards. In April 2022, BLS began using J.D. Power transaction data for new vehicles, and in 2023 it changed how it handles model changeovers (2022 notice, 2023 notice), so the most recent years are not measured exactly as earlier ones were.

What else changed around 2008

The break in the line lines up with several events at once:

• December 2007: EISA became law, requiring a combined average of at least 35 mpg by 2020 on the lab test. The recession began the same month and ended in June 2009 (NBER).
• Model year 2008: footprint-based truck standards began. On the compliance test, the industry as a whole beat its required average by 2.3 mpg (2008), 3.5 mpg (2009) and 3.9 mpg (2010).
• 2009: the Obama administration finalized the 2011 standards in March. Chrysler and GM went through bankruptcy that spring and summer, and "Cash for Clunkers" paid rebates on about 677,000 new-vehicle purchases in July and August (CRS).
• 2007–2011: the yen strengthened sharply against the dollar, raising the dollar cost of Japanese-built vehicles, and the March 2011 earthquake in Japan disrupted production (Federal Reserve).
• Model years 2009–2012: electronic stability control phased in as a federal safety requirement. The joint NHTSA/EPA standards of 2010 began with model year 2012 vehicles.

How CAFE treats electric vehicles

The law bars NHTSA from counting electric vehicles' fuel economy when it sets the standards (49 U.S.C. 32902(h)). EVs still count when automakers comply: DOE's petroleum-equivalency factor converts an EV's electricity use into a gasoline-equivalent rating, and because those ratings are high, EV sales raise an automaker's average and can generate credits to sell to other automakers.

Before 2025, NHTSA's analyses included the EVs that state mandates and market trends were expected to put on the road; its 2024 rule projected battery-electric vehicles at about 28% of new light vehicles by 2031 (89 FR 52737). The 2026 final rule leaves EVs out of the analysis entirely, ends trading of credits earned from 2028, and uses DOE's February 2026 factor, which lowers the CAFE rating of DOE's example EV (244.75 watt-hours per mile) from 335.2 mpg to 50.3 mpg in petroleum-equivalent terms (not the window-sticker MPGe).

The administration argues the earlier standards effectively forced electrification (NHTSA, June 2025); NHTSA's 2026 final rule says they were not attainable for the gasoline- and diesel-powered fleet (p. 25). California and 21 other states' attorneys general argue the law bars only counting EVs' fuel economy when setting standards, not recognizing that EVs are being sold (comment, February 2026). California's own rule does require EV sales; see EV policy by country.

Penalties and credits

Automakers that fall short can use credits earned in other years or bought from other companies. Any remaining shortfall used to mean a civil penalty. Fiat Chrysler (now Stellantis) paid, for example:

• $77.3 million for model year 2016
• $79.4 million for model year 2017
• $123.3 million for model year 2018

Source: NHTSA CAFE civil penalties (July 24, 2025 edition). We could not extract the full payment table, so no totals are shown.

Congress set the CAFE civil penalty to $0.00 in July 2025. The change covers cases in which NHTSA had not yet sent a penalty notice. (P.L. 119-21 §40006) NHTSA's penalty regulation still listed $17 per vehicle for each tenth of a mile per gallon of shortfall for model year 2024 when we checked on September 28, 2026; the statute controls. (49 CFR 578.6)

Tesla's regulatory-credit revenue

Credit sales were a real business for EV makers. Tesla reports one figure for all the regulatory credits it sells to other automakers and does not break it out by program, so these numbers are not CAFE alone:

Annual rows are full years; 2026 rows are single quarters.

Stellantis: net €97 million expense in 2025

• Reversal of its accrued CAFE penalty provision (a gain): €844 million
• Write-down of purchased regulatory credits: €609 million
• Charge for credit-purchase commitments it no longer needs: €504 million
• Net CAFE credits recognized (a gain): €172 million
• Net expense: €97 million

FY2025 annual report (SEC)

General Motors: net $0.5 billion charge in the second quarter of 2026

• Net charge on compliance-related assets, attributed to the April 2026 repeal of EPA's endangerment finding: $500 million
• Carrying amount of compliance-related assets at June 30, 2026: $700 million

Q2 2026 quarterly report (SEC)

Stellantis attributes its 2025 figures to the July 2025 law that set CAFE penalties to $0.00. GM attributes its 2026 charge to EPA's repeal of the greenhouse-gas standards.

Frequently asked questions

What are CAFE standards?

Corporate Average Fuel Economy (CAFE) standards are federal rules, set by NHTSA under a 1975 law, that require each automaker's new cars and light trucks to meet a sales-weighted average fuel economy. Each automaker's target depends on the size (footprint) of the vehicles it sells, so there is no single mpg number every company must hit.

What mpg does the law require now?

Until the new rule takes effect, the standards in force are those of the 2024 rule, which projected 50.4 mpg (projected fleet average, model year 2031, lab compliance test, with AC and off-cycle credits). NHTSA signed the final rule on September 25, 2026 and posted the signed text on its website. As of September 28, 2026, it had not yet been published in the Federal Register; it takes effect 60 days after publication. The final rule projects 34.9 mpg (projected fleet average, model year 2031, lab compliance test, without AC and off-cycle credits) (NHTSA signed final rule (pre-publication text), pp. 26, 54).

Why is CAFE mpg higher than my window sticker?

CAFE uses an older two-cycle lab test. The window sticker uses a five-cycle test meant to reflect real driving, and NHTSA says real-world fuel economy is generally 20 to 30 percent lower than compliance values. EPA's 2024 Prius example: 80 mpg (lab compliance test), 57 mpg (window-sticker rating), 56 mpg (EPA real-world estimate).

Is there an EV mandate?

Technically, no. In practice, the federal standards were built so that automakers could meet them only by selling large numbers of electric vehicles. No federal rule required a set share of EV sales, and NHTSA and EPA each said so in their 2024 rules. But EPA's 2024 greenhouse-gas rule projected electric vehicles, including plug-in hybrids, at 68% of new sales by 2032 (89 FR 28087), and NHTSA's 2024 CAFE rule projected battery-electric vehicles at about 28% of new vehicles by 2031 (89 FR 52737). EPA said most of that shift would come from market forces such as the federal EV tax credits, which ended in September 2025 (89 FR 28057). NHTSA's 2026 final rule concludes the earlier CAFE standards were not attainable for the gasoline- and diesel-powered fleet. California's Advanced Clean Cars II rule does mandate EV sales, rising to 100% of new cars and light trucks by 2035; Congress disapproved its federal waiver in 2025, and that is in litigation.

Do CAFE standards raise car prices?

The price data alone cannot answer that. Relative to overall inflation, BLS's quality-adjusted new-vehicle prices fell about 3.3% a year from 1998 to 2008, then fell about 0.4% a year from 2009 to 2016. Several things changed around 2008 at once, including the recession, the GM and Chrysler bankruptcies and new fuel economy rules. The analysis section of this page summarizes the argument that the standards raised prices and EPA's rebuttal.

What happens if automakers miss the standard?

They can use credits earned in other years or bought from other automakers. Until 2025, any remaining shortfall meant a civil penalty; Fiat Chrysler (now Stellantis), for example, paid penalties for model years 2016 through 2018. Congress set the CAFE civil penalty to $0.00 in July 2025. The change covers cases in which NHTSA had not yet sent a penalty notice.

What did the Trump administration change in 2026?

NHTSA signed the final rule on September 25, 2026 and posted the signed text on its website. As of September 28, 2026, it had not yet been published in the Federal Register; it takes effect 60 days after publication. NHTSA projects the final standards at 34.9 mpg (projected fleet average, model year 2031, lab compliance test, without AC and off-cycle credits), compared with 49.3 mpg (projected fleet average, model year 2031, lab compliance test) under the 2024 standards. The rule, once in effect, revises standards back to model year 2022, ends trading between automakers for credits earned from 2028, and moves many crossovers into the passenger-car class from 2030. From 2028 NHTSA also leaves air-conditioning and off-cycle credits out of the analysis it uses to set the standards; EPA's credit programs are unchanged.

Does the rollback affect gas prices?

Not directly. CAFE governs how efficient new vehicles are, which affects how much fuel drivers buy over many years, not the price of a gallon. NHTSA projects 4.6% more gasoline use through 2050 than under the 2024 standards (NHTSA signed final rule (pre-publication text), p. 55).

Sources and how this page is updated

Every figure on this page is transcribed from, or calculated from, the primary source linked next to it. The BLS price data refresh automatically each month. The EPA and NHTSA tables are updated by hand when those agencies publish new editions, and the rollback section is updated when the final rule reaches the Federal Register.