AAA vs EIA Gas Prices: Why the Averages Differ

EIA’s survey put U.S. regular gasoline at $4.269 a gallon at 8:00 a.m. on Monday, October 26, 2026. AAA’s national average read $4.2258 on Thursday, October 29, and the Bureau of Labor Statistics’ average for regular in September was $4.482, against EIA’s September figure of $4.355. None of these is wrong. EIA prices a sample of 800 to 1,000 stations once a week in cash and weights each by its sales volume; AAA publishes a daily figure from up to 120,000 stations through OPIS; the BLS averages daily prices reported to an outside data source across the whole month.

The gap between the BLS and EIA monthly averages for regular has been between +12.7¢ and +20.5¢ in every one of the 21 months from January 2025 to September 2026, always with the BLS higher. The AAA and EIA numbers, by contrast, sat within a few cents of each other in the October 2026 week compared below, and 0.3 cents apart a year earlier. Which gap you see depends on which pair of publishers you put side by side, and on whether the two figures describe the same day.

Four publishers, four different measurements

Each national figure answers a slightly different question. EIA asks: what was the volume-weighted cash price of self-serve gasoline at 8:00 a.m. local time this Monday? AAA asks, in effect: what is today’s average across as many stations as a commercial price service can see? The BLS asks how much gasoline prices changed from last month for urban consumers, and publishes a dollar average as a by-product. GasBuddy runs a consumer app that shows station prices reported largely by users.

Those questions lead to different samples, different clocks and different arithmetic. A survey that weights by volume gives a high-volume warehouse-club station more pull than a rural single-pump station; a simple average over every station would give them the same pull. A cash price and a card price at a station that posts a cash discount are two different numbers. A Monday-morning snapshot taken in a week when prices are falling will sit above a Thursday daily average from the same week, even if both are measured perfectly. Keep those three mechanics in mind and most headline gaps explain themselves.

The method table, built from each publisher’s own text

Table 1 sets the four methods side by side. Every cell comes from the publisher’s own methodology page, survey form or factsheet. Where the publisher does not state something, the cell says “not published”; nothing has been filled in from third-party descriptions. The one exception is marked: EIA’s own 2016 description of AAA and GasBuddy, quoted as EIA’s account, not as theirs.

EIA is the only publisher that documents its whole recipe, down to the formula for a confidence interval. And the two private publishers say almost nothing about the three variables that most often move a national average by a few cents: weighting, time of day and payment type. That does not make their numbers worse. It means a reader cannot attribute a gap to a specific cause with the same certainty.

How EIA builds the Monday number

The EIA figure that newspapers call “the government average” is a statistical estimate from Form EIA-878, the Motor Gasoline Price Survey. Response is mandatory for the stations selected. Each Monday, the survey collects the price of regular, midgrade and premium as of 8:00 a.m. local time. The instructions to stations are specific: report the cash pump price for self-serve unleaded gasoline; if the station has no self-serve, report mini-serve or full-serve; if it does not accept cash, report the credit card price. Gasoline with more than 15% ethanol is outside the survey.

The sample behind it

The current sample dates from May 14, 2018, when EIA replaced the previous one. It was drawn from a frame of about 130,000 retail outlets active in 2016, assembled from a commercial database, EIA’s own frames, federal and state records and public sources. EIA split the frame into strata by location, matching its published cities, states and regions and the EPA’s reformulated gasoline areas, then split each stratum again into big-box and other outlets. Big-box stations, the warehouse clubs and large discount retailers, were given extra weight in the allocation because each sells more fuel. The result was a stratified systematic sample of 1,000 outlets, which EIA augments each year for new stations and oversamples where closures thin it out.

EIA’s survey forms page lists 800 respondents for Form EIA-878. EIA does not reconcile that figure with the 1,000-outlet design on its methodology page, so treat “about 800 to 1,000 stations” as the honest range. EIA’s January 2016 Today in Energy note, written under the previous sample, gave 802 stations. The methodology page carried on the October 27, 2026 release still describes the 2018 sample as current, and neither the Gasoline and Diesel Fuel Update page nor the survey forms page shows a later notice of a new gasoline sample. The diesel survey is different: it changed its own sample on June 13, 2022, which matters if you compare gasoline and diesel histories.

Volume weights, not a head count

Each sampled station carries two weights multiplied together. The sampling weight is the inverse of the station’s chance of selection. The volume weight is its annual sales of that grade, reported once on Schedule B when the station enters the sample and never published. So a busy warehouse-club station in a suburb counts for more than a corner station selling a tenth as much. If high-volume stations tend to price lower, and big-box outlets are built to do that, a volume-weighted average will sit below a simple average of posted prices from the same stations.

What EIA does when a station does not answer

Missing prices are imputed from two sources: other stations in the sample, and weekly prices from a private commercial source. EIA states that reporting stations account for at least 80% of weighted annual volume for regular at the U.S. level, and it withholds an estimate if half or more of its weighted volume rests on imputed prices. Any estimate whose coefficient of variation exceeds 5% is flagged.

The margin of error nobody quotes

EIA publishes a standard error for every average, and its methodology page gives the recipe for a confidence interval: multiply the standard error by 1.96, round up to the nearest tenth of a cent, and add and subtract. For October 26, 2026, the U.S. regular standard error was $0.0203, which gives a 95% margin of $0.040 and an interval of $4.229 to $4.309. A 2-cent gap between EIA and another publisher on the same day is inside the survey’s own sampling uncertainty. A 15-cent gap is not.

What AAA publishes, and what it leaves out

AAA’s fuel price site says it is “updated daily by Oil Price Information Service (OPIS)” with national, state and local averages for gasoline, diesel and E85, and that “every day up to 120,000 stations are surveyed in cooperation with WEX, Inc.” That is the full public method. AAA’s page does not say what time of day the figure represents, whether it is a cash or card price, whether taxes are in it, or how stations are weighted into the national average.

The only third-party description from a primary source is EIA’s own, from January 2016: AAA’s Daily Fuel Gauge Report is updated daily by OPIS “based on credit card transactions at up to 120,000 stations,” and, like GasBuddy, does “not represent a specific point in time” or separate prices by formulation. That is EIA describing AAA a decade ago, and AAA’s page today names WEX rather than card transactions in general. If AAA’s figure does come from card transactions, it would read a station that posts a cash discount at its card price, while EIA would read the same station at cash. That is a direction, not a measured size; neither publisher states how much it matters.

AAA does publish one adjustment openly: a BTU-adjusted E85 price. Its page notes that E85 delivers about 25% fewer BTUs per gallon than conventional gasoline and that the adjusted figure “is not an actual retail average price paid by consumers.” Readers comparing E85 across sources should check which of the two numbers they are looking at.

GasBuddy: a station map with no published averaging method

GasBuddy’s About page says the app provides “real-time gas price information at 150,000+ stations.” Its site does not publish how the national average quoted in its press commentary is calculated: not the weighting, not the time window, not whether user reports and other inputs are blended. EIA’s 2016 note says GasBuddy “largely relies on user-reported gasoline prices,” and that is the only primary-source description available.

In practice that makes GasBuddy a strong tool for one job, finding the price at specific stations near you, and an unverifiable one for another, comparing its national figure with EIA’s to the cent. The GasBuddy column in Table 1 is mostly “not published” for that reason, and Table 2 has no GasBuddy figure. Claims about how many times a day its survey updates circulate in syndicated press-release copy, not on any GasBuddy methods page, and they cannot be checked against a published procedure.

CPI gasoline: an index first, a dollar average second

The BLS publishes two different gasoline products, and mixing them up is the most common error with this source. The first is the CPI gasoline (all types) index, a pure measure of change on a 1982-84 = 100 base. In September 2026 it stood at 383.182 not seasonally adjusted, and 376.558 seasonally adjusted. Neither is a price. The second is the average price series, in dollars per gallon, for regular, midgrade, premium and all types, published monthly for the U.S. city average and regions. September 2026 regular was $4.482.

Gasoline is no longer priced through the BLS’s own data collection. The BLS motor fuel factsheet says the gasoline index uses a secondary source, takes every observation from it that falls inside CPI geography, and is limited to observations self-reported by that source’s users. Each record is a daily average price per gallon for one fuel at one outlet, with its ZIP code and coordinates, and monthly figures are built from those daily files. Outliers are removed and missing station prices imputed. The factsheet does not name the source and does not say when the switch was made. In the CPI basket, gasoline carried a relative importance of 2.895 as of December 2025, almost all of the 2.981 for motor fuel.

The index and the BLS all-types dollar average move together closely. From August to September 2026 the index rose +6.52% and the all-types average price +6.51%. They are not identical because the index chains county-level geometric means of each station’s price change, while the average is a level. The seasonally adjusted index rose +9.1% over the same month, because September prices normally ease and the adjustment credits the rise as larger than usual. For dollars-and-cents comparisons with EIA or AAA, use the not seasonally adjusted average price series.

One week compared: October 23 to 30, 2026

Table 2 lines up every dated national figure that EIA and AAA show for the last full week of October 2026. EIA’s values are its Monday readings; AAA’s are the daily figures on its site as of October 30, including the “week ago” and “month ago” values it displays. GasBuddy does not publish a dated national figure on its own site, so it has no row. The BLS has no weekly figure.

Read in date order, the two series tell one story. EIA’s Monday reading changed by -1.5¢ from October 19 to October 26. AAA’s Friday figure for October 23, which falls between those Mondays, was $4.2926: +0.9¢ above the earlier EIA reading and +2.4¢ above the later one, and inside both EIA confidence intervals. By Thursday, October 29, AAA had dropped to $4.2258, -4.3¢ against EIA’s Monday price, and AAA’s own release that day put the weekly decline at 5 cents. The AAA figure is lower on Thursday because prices kept falling after Monday morning, not because AAA measures low.

The year-ago pair confirms that the two series are close when the market is calm: $3.035 from EIA on Monday, October 27, 2025, and $3.0377 from AAA on Thursday, October 30, 2025, a difference of 0.27 cents. The month-ago pair, $4.465 on Monday, September 28 against $4.4343 two days later, again shows AAA lower in a falling week.

The grade spread is where the two really differ

Regular prices from the two publishers agree within a few cents, but the premium spread does not: EIA’s premium sat +100.6¢ above its regular on October 26, while AAA’s sat about 91 cents above on both Fridays either side. Because the AAA spread barely moved between October 23 and 30, the 9-to-10-cent difference is not a timing effect. The published methods cannot settle its cause. EIA weights each grade by that grade’s own sales volume at each station; AAA does not say how it weights grades or whether its premium figure covers the same stations as its regular figure. For the reasons the premium gap exists at all, see our premium vs regular price gap analysis.

Formulation hides inside every EIA national figure

EIA also splits the October 26 regular price by formulation: $4.121 in conventional-gasoline areas and $4.587 in reformulated-gasoline areas, a gap of +46.6¢. The national $4.269 blends the two by volume. AAA publishes no such split. The all-grades figure, $4.405, is another trap: it is +13.6¢ above regular because it mixes in midgrade and premium. A headline that compares EIA “all grades” with AAA regular has a built-in 13-cent error before any method question arises. Why reformulated areas cost more is covered in our summer vs winter gasoline guide.

One month compared: September 2026

Monthly figures let the BLS join the comparison. EIA does not run a separate monthly survey; its monthly price is the simple mean of that month’s Monday readings. September 2026 had four Mondays, at $4.157 (Sep 7), $4.319 (Sep 14), $4.478 (Sep 21), $4.465 (Sep 28). Their mean is $4.3547, and EIA’s published monthly figure is $4.355. The same identity holds, to the tenth of a cent, for every month from January 2025 to October 2026. AAA does not publish a monthly national average on its site.

All six series agree on direction and broadly on size. EIA’s regular rose the most, +7.3%, against +6.7% for the BLS regular average. With only four or five readings a month, EIA’s average depends on where the Mondays fall in a month of fast change, while the BLS average takes in every day. In levels, the BLS regular average sits +12.7¢ above EIA and the BLS all-types average +14.0¢ above EIA’s all grades.

The BLS average runs above EIA every month

One month could be noise. Table 5 runs the comparison for 21 months. The BLS regular average was higher than EIA’s regular every month, by +12.7¢ at the narrowest (September 2026) to +20.5¢ at the widest (March 2026), with a mean of +15.8¢. The month-to-month changes, by contrast, stay within about 2 percentage points of each other, which is what you would expect if both series track the same market from different starting points.

A gap this persistent points to method, not chance. Table 1 lists the documented differences that push in a consistent direction, and the published texts allow no more than that list:

• Volume weighting. EIA weights each station by its sales and gave big-box outlets extra allocation in its sample design. The BLS factsheet does not describe how its dollar averages weight stations.
• Cash price. EIA records cash wherever a station accepts it. The BLS factsheet does not state a payment basis for its gasoline observations.
• Who reports. EIA’s prices come from a probability sample of stations required to respond. The BLS uses observations self-reported by users of an outside source, which reflects the stations those users visit.
• Geography. The BLS figure is a U.S. city average for urban consumers; EIA’s sample covers all U.S. retail outlets in its frame.

Which of these explains most of the 13 to 21 cents cannot be read off the published methods, and this page does not assign shares. The practical rule is simpler: never subtract an EIA figure from a BLS figure, or the reverse, and call the result a price change.

The widest gap, March 2026, came in the month of the sharpest move in the table. EIA’s weekly regular went from $3.015 on March 2 to $3.990 on March 30, and the month’s average rose +25.1%. When prices climb that fast, a series of five Monday snapshots and a series of 31 daily averages can land a few cents apart on timing alone. The two series still agreed on the size of the jump within half a point.

Which number fits which job

The right series depends on what the number is for. Table 6 is a decision guide, built from the method differences above rather than from any view of which publisher is “best.”

The state coverage limit in the last row matters more than it looks. EIA’s weekly table publishes California, Colorado, Florida, Massachusetts, Minnesota, New York, Ohio, Texas and Washington, plus Boston, Chicago, Cleveland, Denver, Houston, Los Angeles, Miami, New York City, San Francisco and Seattle. For any other state, AAA is the readily available daily source, and an AAA figure for, say, Arizona should be compared with AAA’s national figure, not with EIA’s. How much of a state gap is tax, supply and rules is broken down in why gas prices differ by state.

Headline traps, and how to spot them

“Prices fell X cents” when two publishers are mixed

A story that quotes last week’s EIA figure and today’s AAA figure reports a change that blends a real price move with the method gap. On October 29, 2026, AAA’s release said the national average fell 5 cents over the past week. EIA’s comparable change, Monday to Monday, was -1.5¢. Both are right for their own window. Comparing AAA on October 29 with EIA on October 19 would give -5.8¢, which is neither.

Record highs that disagree

Each publisher has its own record, set on its own day. A daily series will usually show a higher peak than a weekly series, because a weekly survey can miss the exact top. An EIA monthly average will show a lower peak still, since it averages four or five Mondays. When two outlets cite different record prices for the same episode, check whether one is quoting a daily, a weekly or a monthly series before assuming either is mistaken.

Holiday weeks and late releases

When Monday is a federal holiday, EIA collects on Tuesday and publishes on Wednesday. Its methodology page says the release still represents Monday’s prices; the survey instructions say Tuesday collection. Either way, the reading is a day later than usual in a holiday week, so a Tuesday collection after a holiday weekend should not be read as an exact Monday-to-Monday change. The CPI arrives with a longer lag: the October 2026 figures are scheduled for November 10, 2026.

Index levels quoted as prices

A CPI gasoline index of 383.2 is not $3.83 a gallon or any other price; it means gasoline cost about 3.8 times its 1982-84 average. The ratio between the index and the BLS all-types dollar price drifted only between about 81.8 and 82.9 across the 20 months in Table 5 that have both figures, so the two series tell the same story in different units. Use the index for percentages and the average price series for dollars.

The pump price itself splits into crude, refining, distribution and taxes, and those shares are the same whichever publisher measures the total; our gasoline price breakdown works through the slices with EIA’s own data. Why a falling crude price takes longer to reach every one of these averages than a rising one is the subject of rockets and feathers in gasoline prices.

Putting a long history on one footing

For history, the EIA monthly and annual regular series are the usual base, and the BLS average price series is the alternative when an older start or an urban-only measure is wanted. What breaks a history is switching between them mid-series: on the 21 months above, a switch from EIA to BLS would add a phantom 13 to 21 cents in one step. The EIA series has its own seam at May 14, 2018, when the current sample replaced the old one; EIA published both samples’ estimates for April 30 and May 7, 2018, so the size of that break can be checked.

To restate a price in today’s money, multiply by the ratio of the CPI-U all-items index in the target month to the index in the original month. That is the method used in our table of gas prices by year, adjusted for inflation, and the same arithmetic runs in the CalculatorVillage inflation calculator. When you see a weekly price in the news and want to know where it sits against the latest EIA reading, the weekly figures are on our gas prices page, and the day each EIA release lands is set out in the EIA report schedule guide. A similar two-publisher puzzle exists for crude stocks, where an industry survey and the government survey disagree every Tuesday and Wednesday; the API vs EIA inventory comparison walks through that one.

Data and sources: U.S. Energy Information Administration, Methodology for EIA Weekly Retail Gasoline Price Estimates (as carried with the October 27, 2026 release), Gasoline and Diesel Fuel Update (release of October 27, 2026), recent gasoline prices and standard errors, weekly and monthly series U.S. Regular All Formulations Retail Gasoline Prices and U.S. All Grades All Formulations Retail Gasoline Prices, Form EIA-878 instructions and the EIA survey forms page, all checked October 31, 2026; EIA, Today in Energy, January 2016 (survey of 802 stations under the former sample; descriptions of AAA and GasBuddy), checked October 31, 2026. AAA, national average gas prices (price as of October 30, 2026), About page and newsroom release of October 29, 2026, checked October 31, 2026. GasBuddy, About page, checked October 31, 2026. U.S. Bureau of Labor Statistics, Measuring Price Change in the CPI: Motor fuel (modified May 20, 2026), CPI release schedule, and series CUUR0000SETB01, CUSR0000SETB01, APU000074714 and APU00007471A, data extracted October 31, 2026. Differences, spreads, intervals, monthly means and percentage changes are this page’s arithmetic.