Why Premium Gas Costs So Much More Than Regular

Premium averaged $4.030 a gallon across the United States in 2025, against $3.097 for regular: a gap of 93.3 cents, or 30.1% more per gallon, in EIA’s annual retail averages. In 2010 the same gap was 24.0 cents. Part of any premium price is real: octane comes from refinery streams that cost more to make than ordinary gasoline. But the widening that began in 2010 is a different story: an EIA-commissioned study of 2010 to 2017 found that it did not come from refining costs, which actually fell over that stretch.

From 1995 to 2009 the gap barely moved, staying between 17.9 and 23.9 cents a gallon whether pump prices rose or fell. Since 2010 it has widened in every single year, through the 2015 crude collapse, the 2020 pandemic slump and the 2022 price spike alike. For a driver, the practical question is narrower: whether the owner’s manual says premium is required or merely recommended, because only the first makes the gap unavoidable. For the four slices inside any gallon, start with what you pay for in a gallon of gasoline; for current local prices by grade, see the gas price map, which is refreshed from EIA data.

The grade gap, year by year, from EIA’s retail survey

EIA publishes three national annual series for retail gasoline by grade, each covering all formulations (conventional and reformulated): regular, midgrade and premium, in dollars per gallon including all taxes. Table 1 takes those published averages and does the subtraction. Regular’s annual series reaches back to 1992, but EIA’s midgrade and premium annual averages begin in 1995. The premium monthly series has its first value in December 1994, so 1994 has no annual average for the upper grades and no gap can be computed for it.

The 2020 gap of 66.6 cents and the 2025 gap of 93.3 cents both fall straight out of these published averages. Two figures from EIA’s own octane study line up with the table too. The study put the 2017 gap at $0.50 a gallon, and Table 1 gives 50.2 cents. It said the spread roughly doubled from about $0.25 in 2010, where Table 1 has 24.0 cents, and that across the fifteen years from 1995 to 2010 it grew by only about $0.05, which matches the move from 18.7 to 24.0 cents.

Three eras in one spread

1995 to 2009: a fixed markup in cents

For fifteen years, premium carried what looked like a flat surcharge of roughly 18 to 24 cents. Because the cents hardly moved while the base price more than doubled, the percentage fell as the base price rose: premium cost 16.8% more than regular in 1995 and only 7.4% more in 2008. (The low point of the whole series came just after, at 7.1% in 2011.) In the years regular got cheaper (1998, 2001, 2002 and 2009), the gap shifted by no more than three-tenths of a cent. A driver comparing grades in that period was looking at a premium of about two dimes, no matter what crude oil was doing.

2010 to 2017: the doubling

The gap went from 24.0 cents in 2010 to 50.2 cents in 2017. The sharpest steps came as pump prices fell: when regular dropped 92.9 cents between the 2014 and 2015 annual averages, the gap grew 4.9 cents, and in 2016 it grew another 7.2 cents while regular fell again. That pattern, premium holding up as regular slides, is the one EIA’s octane study set out to explain, and it is covered in detail below.

2018 to 2025: past 90 cents

The gap kept climbing after the study period, to 57.5 cents in 2018, 66.6 in 2020, 79.3 in 2022 and 93.3 in 2025. As a share of the regular price it reached 30.7% in 2020, the highest of the series, and 30.1% in 2025. Since 2010 regular has fallen year over year 9 times (2013, 2014, 2015, 2016, 2019, 2020, 2023, 2024, 2025), and the gap widened in every one of those years. It also widened in each year that regular rose. Sixteen consecutive annual increases is the plainest fact in the table, and it is not something a cost-of-octane story alone would predict.

Why octane costs anything to make

The number on the yellow pump label is an anti-knock index: the average of the research octane number and the motor octane number, written (R+M)/2. EIA describes the higher number as a more stable fuel, one that resists igniting on its own under compression, and notes that the label shows a minimum. EIA’s survey definitions put regular at an index of at least 85 and below 88, midgrade at 88 to 90, and premium above 90. Retail premium is generally 91 to 94, according to both EIA and fueleconomy.gov.

Finished gasoline is a blend of refinery streams, and those streams differ widely in octane. The table below is condensed from the EIA octane study’s summary of refinery blendstocks. Raising the octane of the pool means using more of the high-octane streams, or running the units that make them harder.

The study calls alkylate, reformate and normal butane the highest-octane components a refinery makes, and it describes the reformer as the traditional swing source of octane: it converts low-octane naphtha into high-octane reformate, and running it at higher severity raises the octane of what comes out. Butane is cheap octane but evaporates easily, which is why its use shrinks every summer; the summer vs winter gasoline guide follows that trade-off through the seasonal price. Outside the refinery, ethanol adds octane too. Fueleconomy.gov gives it an octane rating of about 109, and most U.S. gasoline contains up to 10% ethanol; the E15 explainer covers why an ethanol share and an octane grade are separate labels.

How much does that extra octane cost at the refinery? The official consumer answer, from fueleconomy.gov, is one sentence: the components that boost octane are generally more expensive to produce. It gives no figure. EIA’s Energy Explained pages describe octane at length but do not price it either. The best measure in the sources reviewed here is indirect. The octane study treats the premium-to-regular spread in the Gulf Coast spot market as the market’s reading of what it costs a refiner to make premium rather than regular, and it expected the New York Harbor spread to track refinery economics as well because that market is well connected to the Atlantic Basin. On that measure, the New York Harbor spot spread rose from about $0.10 to roughly $0.20 a gallon between September 2010 and May 2011, peaked near $0.37 in September 2012, then trended down to an average of $0.14 in 2016. In Houston it stayed under $0.20 for most of 2016. So the refinery cost of the upgrade, read through spot markets, was in the tens of cents while it lasted, and it was falling while the retail gap rose.

What EIA’s octane study found behind the widening

In 2018 EIA published a two-phase study by the consultancy Baker & O’Brien, commissioned specifically to find out whether changes in refinery operations explained the widening gap. Its scope was refining. It states plainly that it is not an examination of retail pricing dynamics. Its answer was no: the rising retail spread was not caused by a higher cost to make high-octane gasoline at U.S. refineries.

The reasoning runs through shale oil. Since 2010 U.S. refiners have run more light, sweet domestic crude, which yields more naphtha and other low-octane blendstocks. Light sweet crude processing more than doubled between 2010 and 2016, and the share of naphtha, natural gasoline and other light, low-octane components in the gasoline pool rose by more than 80%. Cheap, abundant naphtha is exactly what reformers use as feed, so refiners ran their reformers more and slightly harder, turning the surplus into high-octane reformate. Capacity was not tight: reformer utilization in 2016 ranged from 73% to 80%. The study found that refiner costs to make premium from regular and reformate had declined since 2014, closely matching the spot spread, and that spot spreads had been narrowing since 2013 apart from brief spikes during temporary problems such as refinery outages. At the rack, the terminal where retailers load trucks, it found no significant change since 2010 that would account for the retail trend.

The study found two other candidates unlikely to matter much. Petroleum-based gasoline consumption in 2016 was still below its 2007 peak once ethanol blending was accounted for, so it found no major shift in demand. And although automakers have raised compression ratios with turbochargers, engine design trends had not yet significantly changed the share of vehicles needing premium. About 10% of light-duty vehicles had owner’s manuals that required or recommended premium; that share was 11.5% to 11.7% of the fleet from 2010 to 2014 and slipped to 9.6% by 2016. It also looked abroad and found no similar widening in other countries’ markets.

Regional patterns pointed the same way. In 2017 the gap averaged $0.56 a gallon in the Midwest (PADD 2), up $0.34 from 2010, while on the West Coast (PADD 5) it rose only $0.08, from $0.23 to $0.31. The two explanations the study offered for the marketing layer are explicitly outside its scope and were not measured. The first is a variant of the rockets and feathers pattern, in which both grades rise together but premium falls more slowly than regular. The second is search cost: buyers can find regular prices more easily than premium prices, so retailers face less pressure to cut premium. As supporting evidence, the study cites Los Angeles, where retailers must post the prices of every grade and where the premium-to-regular spread barely changed over the study period. It also observed that premium’s share of sales rose as absolute prices fell from 2014, which it read as a sign that the marginal premium buyer responds to the price of premium itself more than to its gap over regular, making premium a separate market. None of this was quantified as a share of the gap, and the study does not claim it was.

2018 to 2025: the part no source here explains

The study’s data end in 2017. The gap has since risen another 43.1 cents, from 50.2 to 93.3. No EIA analysis reviewed for this article examines that later period, so its causes are not established here, and carrying the 2018 study’s conclusion forward would be a guess presented as a finding. What the data alone can show is limited but useful. The widening continued in years when regular rose, such as 1.3 cents in 2021 and 11.4 cents in 2022, and in years when it fell, such as 4.2 cents in 2025. That rules out any simple story in which the gap only opens on the way down.

Premium also remains a small part of the market. In EIA’s refiner sales volumes, 2019, the last full pre-pandemic year, shows premium at 2,495.6 thousand gallons a day out of 24,239.3 thousand gallons a day in sales to end users, about 10.3%, with midgrade at about 6.7%. That series covers refiners’ own sales to end users, not every station, and its published annual history ends in 2021. A small, less price-shopped grade fits the search-cost explanation, but fit is not proof, and the shares say nothing about margins by themselves.

Midgrade now sits closer to premium than to regular

The middle grade tells its own story. In 2010 the step from regular to midgrade and the step from midgrade to premium were both 12.0 cents. By 2025 the first step was 57.4 cents and the second only 35.9 cents, so midgrade cost more than half of the way to premium. The lower step has been the larger of the two in every year since 2014. If regular is 87, midgrade 89 and premium 91, the 2025 averages work out to about 28.7 cents per octane point for the first two points and about 18 cents per point for the next two. That per-point arithmetic is an illustration resting on assumed grades: EIA defines midgrade as 88 to 90 and premium as anything above 90, and many stations sell 93.

For a driver whose manual calls for 89, this matters more than the premium gap. Midgrade has not been the cheap compromise its octane number suggests for several years, and the published averages offer no reason to expect the pattern to reverse on its own. No EIA source fetched for this article explains the faster growth of the lower step.

What premium costs one driver over a year

Take a car driven 12,000 miles a year that averages 25 miles per gallon. Those are illustration inputs, not national averages from a cited source. The car burns 480 gallons a year. At EIA’s 2025 annual averages, filling with regular costs 480 × $3.097 = $1,486.56, and filling with premium costs 480 × $4.030 = $1,934.40. The difference is $447.84 a year, which is the same as 480 gallons times the 93.3-cent gap. Over ten years at that gap, with no discounting, it adds up to about $4,478.

The same driver, buying the same 480 gallons, would have paid about $94 a year extra at the 1995 to 2009 average gap of 19.6 cents, and $320 at the 2020 gap. In nominal dollars the cost of the habit has roughly quintupled, though the fuel itself has not changed. Table 4 shows how the annual extra moves with fuel economy and with the size of the gap, at 12,000 miles.

A quicker test works for any year. Premium only pays for itself on fuel cost if it raises your miles per gallon by the same percentage that it raises the price per gallon. In 2025 that was 30.1%: a car getting 25 mpg on regular would need about 32.5 mpg on premium just to break even. In 2008, when the gap was 7.4% of the price, the hurdle was far lower. The percentage column of Table 1 is that hurdle for every year since 1995. To run the numbers with your own mileage and prices, the fuel cost calculator on CalculatorVillage handles the gallons and dollars; enter the regular and premium price as two separate runs.

Required or recommended: a rule that follows the manual

The decision turns on one word in the owner’s manual. Fueleconomy.gov’s instruction is to use the octane rating the manufacturer requires, and it separates two cases: a vehicle that requires midgrade or premium, where higher-octane fuel is worth the extra cost, and a vehicle whose manual does not require premium but says it will run better on higher octane, where the choice is the owner’s and the cost increase is typically higher than the fuel savings. The octane study adds a detail that explains the difference. EPA allows manufacturers to certify fuel economy on either 87 or 91 gasoline, and premium-recommended vehicles are certified on 91 but do not require it for daily driving. The mileage on such a car’s label therefore reflects premium fuel.

Two cases sit outside the table. A car that requires premium but gets a tank of regular in an emergency is not in the same position as one run on regular for years; the sources describe harm over time, and the manufacturer is the authority on any single event. And knock that persists on the grade the manual calls for is a mechanical question, not a fuel-grade one, because the fuel already meets the specified octane; EIA notes that the number on the pump label is a minimum. For a car designed for 87, the premium gap in Table 1 is close to pure cost. For one that merely recommends premium, it is a trade between price and some power or efficiency that only the owner can value. For one that requires it, the gap is simply part of the car’s running cost, and it belongs in the purchase decision, since at 2025 averages it was worth about $448 a year for the illustration driver above.

What the national average does and does not measure

Every Monday, EIA collects prices for all three grades from a sample of roughly 1,000 stations as of 8:00 a.m., mostly by phone and email, and turns them into volume-weighted averages for the nation, regions and selected states and cities. The prices include all taxes and are self-serve cash prices except where only full service exists. The annual figures in Table 1 are averages of that weekly survey. EIA changed its statistical method for weekly retail gasoline estimates on May 14, 2018, and notes that estimates from that week are not directly comparable with the week before; annual averages smooth this but do not erase it.

Three limits follow. A national gap hides wide regional differences, as the study’s PADD figures show, so a local gap can be much larger or smaller; the guide to why gas prices differ by state covers the taxes and fuel specifications behind that. Cash prices miss card surcharges and loyalty discounts. And a cents gap is nominal; for what a gallon of regular cost in constant dollars over the decades, see gas prices by year, adjusted for inflation. The percentage column in Table 1 is the inflation-free way to compare the premium across years, since it compares two prices from the same year.

At the pump itself, the octane number is a regulated disclosure. Under the Federal Trade Commission’s fuel rating rule, 16 CFR Part 306, the octane rating is defined as the research octane number plus the motor octane number, divided by two; refiners, importers and producers must determine the rating before they transfer fuel and certify it to the next buyer; and a retailer must post the rating of all automotive fuel it sells, with at least one label on each face of each dispenser. The label tells you the octane you are paying for. The widening of the price gap above it, as the study found for 2010 to 2017, happened after the fuel left the refinery.

Before the next fill-up decision, the most useful single link is the gas price map: put your local regular and premium prices into the break-even test above, because a local spread can differ a lot from the 93.3-cent national figure.

Data and sources: U.S. Energy Information Administration, retail gasoline prices by grade, all formulations, annual: regular, midgrade and premium, release of October 27, 2026, checked October 30, 2026; and the monthly premium series (first value December 1994), checked October 30, 2026. EIA, retail price definitions, sources and notes, checked October 30, 2026 (grade definitions by anti-knock index, survey method, May 14, 2018 methodology change). EIA, Analysis of Octane Costs, November 28, 2018, and its Phase 1 report prepared by Baker & O’Brien, checked October 30, 2026 (2010 and 2017 spreads, PADD spreads, reformer utilization and severity, crude slate change, spot and consumer spreads in the Central Atlantic and Houston, the 61% marketing share, fleet shares, EPA certification fuels, Table 2 blendstock octane, the rockets-and-feathers and search-cost hypotheses and the Los Angeles posting evidence). EIA, Gasoline explained: octane in depth, checked October 30, 2026 (anti-knock index, grade ranges, minimum on the label, knock sensors and timing). EIA, U.S. Sales to End Users, Total Refiner Motor Gasoline Sales Volumes, checked October 30, 2026 (2019 sales by grade). U.S. Department of Energy and EPA, Selecting the Right Octane Fuel, checked October 30, 2026. eCFR, 16 CFR Part 306, Automotive Fuel Ratings, Certification and Posting, checked October 30, 2026. Gaps, percentages, era averages, the break-even test and the worked example are this page’s arithmetic; the mileage and mpg inputs are illustrations.