1. The one report that moves oil every week
Every Wednesday at 10:30 a.m. Eastern, the US Energy Information Administration publishes the Weekly Petroleum Status Report. It is the fastest official look at the physical side of the US oil market: how much crude sits in tanks, how hard refineries are running, how much fuel the country is actually using, and where spot prices closed the prior week. Futures traders react within seconds, and headlines reduce the whole release to a single word - build or draw. That shorthand is where most readers get misled.
A build means inventories rose during the survey week. A draw means they fell. Neither word, on its own, tells you whether the market is tight or loose. A two-million-barrel draw during peak driving season can be exactly what the calendar predicts, while a small build in a week when everyone expected a draw can tighten the prompt market more than the headline suggests. Reading the report well means reading it in layers: first the stock change, then the flows that caused it, then where stocks sit against their own recent history.
This guide uses PetroEyes archived EIA series - the same weekly data behind our EIA inventory terminal - to show the reading order that professionals use. Every number in the tables below is computed at build time from that data file, with the week-ending date shown, so you can check the math against the source rather than taking a summary on trust. The EIA itself publishes the report free at eia.gov, in the Petroleum Supply Weekly section. Nothing in this guide requires a paid terminal.
2. Know the calendar before you read a single number
The report describes a week that ends on Friday and is published the following Wednesday. That five-day lag creates the first common mistake: treating the release as news about today. It is news about last week, released into a market that has already traded four sessions since the survey closed. When prices jump on the print, they are usually reacting to the gap between the reported number and what analysts expected, not to the physical change alone.
Two calendar traps deserve a permanent note in your notebook. Holiday weeks shift the release. The EIA's explanatory notes say that for weeks which include holidays, releases are delayed by one day, and its published exceptions for 2026 are all Thursday releases at noon eastern, mostly after a Monday holiday. Thanksgiving is the holiday people expect to cause a delay, and it does not: in the EIA archive, the Thanksgiving-week report came out on the Wednesday in every year from 2015 to 2025. Our release schedule guide counts every shift since 2015. Storm weeks are the second trap. A Gulf Coast hurricane that shuts refineries and ports in the same week can produce a huge crude build (refineries are not taking barrels) next to a huge product draw (refineries are not making fuel). Read as a pair, that combination says disruption, not weak demand. Read separately, each half tells a false story.
The report is also preliminary. Production is modelled rather than surveyed, exports are estimated from weekly customs filings, and imports come from a weekly sample of importers scaled up to the full industry. The EIA does not go back and quietly rework old weeks: its stated policy is to disseminate a revised weekly figure only when the revision is expected to substantively affect understanding of US supplies, in the next scheduled release. The fuller record arrives later, as a separate number, in the monthly Petroleum Supply Monthly. A serious reader therefore treats any single week as an estimate with error bars, and gives the four-week average more weight than the latest point for every flow series. The stock levels themselves - actual tank measurements reported by operators - are the hardest data in the release. The flows are the softest. That hierarchy should shape how confidently you argue from each block.
3. The five blocks, in the order to read them
The status report is organised into prices, inventories, production and trade, refining, and demand. Most commentary reads them in that printed order. A better order is stocks first, refining second, demand third, supply and trade fourth, and prices last - because prices are the result, not the cause. Here is what each block actually measures, in plain terms.
Inventories: what is in the tanks
Inventory is reported in thousand barrels, shown in our data as MBbls. Crude oil stocks exclude the Strategic Petroleum Reserve, which the report tracks on its own line for exactly that reason: commercial tanks answer to the market, while the SPR answers to the Department of Energy. Gasoline and distillate stocks are finished products plus blending components in the gasoline case, and distillate covers diesel and heating oil together, which is why winter distillate draws often say more about heating demand in the Northeast than about trucking.
Refining: what the system is doing with crude
Gross refinery inputs tell you how much crude and other feedstocks refineries processed each day, in thousand barrels per day. Refinery utilization expresses those runs as a percentage of operable capacity. The two series answer different questions. Inputs can fall because a plant is down for maintenance even while national utilization looks respectable, and utilization can dip in early autumn simply because turnaround season has started on schedule. Our explainer on refinery crack spreads shows why runs respond to margins: when the crack is wide, refiners push utilization toward its practical ceiling, and the status report is where you verify that they actually did. The ratio itself, and why one PADD can print above 100 percent, is recomputed week by week in the refinery utilization guide.
Demand: product supplied, not pump sales
The demand block is widely misunderstood. The EIA does not survey drivers at filling stations each week. It calculates product supplied - the volume that disappears from primary storage into the distribution system - and uses it as a proxy for consumption. Over a month, product supplied tracks real demand well. In a single week it can swing on pipeline scheduling, terminal transfers, and reporting timing. That is why the guide below insists on the four-week average for gasoline and distillate supplied before calling any demand trend.
Production and trade: the softest numbers
Domestic production in the weekly report is a modelled estimate built from the EIA's Short-Term Energy Outlook, and the EIA may re-benchmark it in the weeks the outlook is released when it drifts from the monthly survey. Treat a 100,000 barrel-per-day weekly wiggle as noise until the monthly data confirms it. Imports come from the weekly importer survey, grossed up with a smoothed ratio because company imports swing so much week to week. Exports are estimated from weekly US Customs and Border Protection filings, with a statistical model applied to exports to Canada where needed, and export weeks are notoriously lumpy because tanker loadings bunch together. One week of exports tells you about shipping schedules. Four weeks tell you about trade.
Prices: the scoreboard
The price block records weekly spot prices for WTI and Brent crude, New York Harbor gasoline, heating oil, and Henry Hub natural gas. These are the same benchmarks we track on the WTI price page and, for gas in depth, on the natural gas price page. In the status report, prices are best used as a check on your physical reading. If stocks drew, runs were high, and product supplied was firm, a falling spot price is telling you the market expected even more tightness - or that the driver sits outside the United States entirely.
4. The latest week, laid out in full
The table below is the original working asset for this guide. It is computed directly from PetroEyes archived EIA data for the week ending November 6, 2026, the most recent week in the file. Columns show the latest level, the prior week, the week-on-week change in level and percent, the four-week average, and the 52-week range, so that any single print can be judged against its own context without opening a second document.
| Series | Unit | Latest | Prior week | WoW change | 4-wk avg | 52-wk range |
|---|---|---|---|---|---|---|
| Crude oil stocks (excl. SPR) | MBbls | 414,417 (2026-11-06) | 409,760 | +4,657 (+1.14%) | 411,830 | 404,508 to 465,729 |
| Gasoline stocks | MBbls | 210,757 (2026-11-06) | 208,958 | +1,799 (+0.86%) | 208,551 | 204,362 to 259,058 |
| Distillate stocks | MBbls | 103,330 (2026-11-06) | 99,259 | +4,071 (+4.10%) | 102,425 | 99,259 to 132,921 |
| Cushing, Oklahoma crude stocks | MBbls | 27,356 (2026-11-06) | 27,250 | +106 (+0.39%) | 27,293 | 18,599 to 31,489 |
| Strategic Petroleum Reserve | MBbls | 279,535 (2026-11-06) | 282,184 | -2,649 (-0.94%) | 281,746 | 279,535 to 415,442 |
| US crude production | MBbls/d | 14,013 (2026-11-06) | 14,003 | +10 (+0.07%) | 13,819 | 13,215 to 14,013 |
| Gross refinery inputs | MBPD | 17,178 (2026-11-06) | 17,054 | +124 (+0.73%) | 16,960 | 15,963 to 17,661 |
| Refinery utilization | % | 95.2 (2026-11-06) | 94.5 | +0.7 (+0.74%) | 94.0 | 88.6 to 98.0 |
| Total product supplied | MBbls/d | 19,759 (2026-11-06) | 21,962 | -2,203 (-10.03%) | 20,790 | 19,226 to 21,962 |
| Gasoline supplied | MBbls/d | 8,767 (2026-11-06) | 8,808 | -41 (-0.47%) | 8,729 | 7,834 to 9,256 |
| Distillate supplied | MBbls/d | 3,411 (2026-11-06) | 4,439 | -1,028 (-23.16%) | 3,947 | 3,156 to 4,753 |
| WTI crude spot price | $/bbl | 95.97 (2026-11-06) | 105.59 | -9.62 (-9.11%) | 102.17 | 56.30 to 107.17 |
Read the current print the way the table invites you to. Crude stocks stand at 414,417 thousand barrels for the week ending November 6, 2026, a week-on-week move of +4,657 thousand barrels. Set against a 52-week range that runs from 404,508 to 465,729 thousand barrels, that move equals about 8 percent of the full range, which is the honest way to size it before calling it large or small. Gasoline stocks built by 1,799 thousand barrels on the week and distillate stocks built by 4,071 thousand barrels, while refinery utilization rose 0.7 points to 95.2 percent. Total product supplied fell on the week. This paragraph is computed from the same archived data as the table every time the page is built, so the reading always matches the latest week in the file. That three-block cross-check, done in under a minute, is the entire craft in miniature: never let one block file the report alone.
5. Days of supply: the worked example that turns levels into tightness
A stock level means little until you divide it by the rate at which the system burns through it. Days of supply does that division. It answers a practical question: if nothing new arrived, how many days would current tanks cover current use? The arithmetic uses only numbers from the table above, all from the week ending November 6, 2026.
Crude days of supply = crude stocks / gross refinery inputs = 414,417 / 17,178 = 24.1 days
Gasoline days of supply = gasoline stocks / gasoline supplied = 210,757 / 8,767 = 24.0 days
Distillate days of supply = distillate stocks / distillate supplied = 103,330 / 3,411 = 30.3 days
These are hypothetical cover calculations, labelled as such: refineries do not actually run tanks to zero, imports keep arriving, and product supplied is a weekly proxy rather than metered consumption. The value of the calculation is comparative, not literal. When gasoline cover falls toward the low end of its recent range heading into driving season, the market has less cushion for an unplanned outage, and the crack spread usually prices that cushion before the status report prints it. When cover rises through autumn as demand fades and refineries return from maintenance, the same arithmetic explains why a headline draw can still leave the market comfortable. Pair days of supply with the 52-week range columns in the table and you have a defensible tightness statement built entirely from public data.
One warning belongs with this tool. Days of supply uses national totals, and the United States is five PADD districts with limited pipelines between them. A comfortable national gasoline number can coexist with a tight East Coast if the surplus sits on the Gulf Coast. The EIA report breaks stocks and runs down by district precisely for this reason. Read the national number for the trend, then check the district where the price benchmark you care about is set.
6. Cushing: the small tank farm with an outsized vote
Inside the crude stock total sits one line that futures traders watch more closely than the national figure: crude stored at Cushing, Oklahoma. Cushing is the delivery point for the NYMEX WTI futures contract, so barrels there are the barrels that can be delivered against the benchmark. When Cushing drains toward operational lows, the prompt WTI contract can strengthen sharply against both later months and Brent, because shorts need deliverable supply. When Cushing refills, that pressure fades even if total US stocks are falling elsewhere.
In the latest week, Cushing held 27,356 thousand barrels, against a 52-week range of 18,599 to 31,489 thousand barrels in our archived series. Placing the print inside that range tells you more than the week-on-week change alone. Our deep dive on the WTI vs Brent spread works through the full mechanism, including how pipeline flows between the Permian, the Gulf Coast, and Cushing move the spread. The weekly habit is simple: read total crude for the balance story, then read Cushing for the benchmark story, and do not confuse the two.
The SPR deserves the same separation discipline. A release from the reserve raises measured supply without any change in commercial production, and a refill does the opposite. Because the status report keeps the SPR on its own line - 279,535 thousand barrels in the latest week - you can strip it out in one glance. Commentary that folds SPR moves into commercial builds or draws is mixing a policy decision into a market signal, and the report format gives you no excuse to join it.
7. Expectations, seasonality, and the four-week rule
Markets trade the surprise, not the level. A widely followed ritual on Tuesday evening, when the American Petroleum Institute publishes its private estimate, and again on Wednesday morning, when analyst surveys circulate, sets the expected build or draw before the EIA prints. The investable information on Wednesday is the distance between the EIA number and that expectation, plus whether the internals agree. A bullish headline with bearish internals - say, a crude draw produced entirely by a one-week export surge while refinery runs fell and gasoline built - tends to fade, because the flows say the draw will not repeat.
Seasonality is the second filter. Every stock series has a calendar, and a print has to be judged against what that week of the year usually delivers, not against zero. A gasoline draw in January is information, because January weeks usually build: 89 percent of January weeks from 2015 to 2024 did. The 52-week range in our table is a rough seasonal scaffold, and the EIA report itself prints five-year averages for exactly this comparison. The calendar is easy to describe from memory and easy to get wrong. Section 8 measures it.
Finally, apply the four-week rule to every flow: production, imports, exports, refinery inputs, and product supplied. Weekly flows carry survey noise, weather noise, and shipping bunching. The four-week average, shown in the table for each series, strips most of that out while still responding within a month when a real trend starts. If the latest week and the four-week average point in opposite directions, believe the average until a second week confirms the turn. More bad Wednesday trades come from ignoring that sentence than from any other single habit.
8. The seasonal calendar, measured: ten years of weekly changes
A common description of the calendar says gasoline stocks build in winter, draw hard through the summer driving season and rebuild in autumn, while distillate builds in summer and draws in winter. We tested that against the EIA's own weekly history. For every week from 2015 through 2024 we took the change in commercial crude, total gasoline and distillate stocks, grouped the changes by the month of the week-ending date, and averaged them. That is 522 weekly changes per product, ten full years, from the historical file behind the report's Table 1. The result keeps half of the rule and breaks the other half.
The average week, by month: US stock changes, 2015 to 2024
Mean weekly change in million barrels, grouped by the month of the week-ending date. Bars above the line are builds.
- Average week builds
- Average week draws
Data behind this chart
| Month | Crude, M bbl | Gasoline, M bbl | Distillate, M bbl | Gasoline weeks that built |
|---|---|---|---|---|
| Jan | +1.5 | +3.9 | +1.3 | 89% |
| Feb | +3.6 | -0.5 | -1.4 | 43% |
| Mar | +2.2 | -2.9 | -1.3 | 11% |
| Apr | +2.3 | -0.1 | -0.4 | 56% |
| May | -0.8 | -0.6 | +0.6 | 32% |
| Jun | -2.3 | +0.3 | +0.5 | 56% |
| Jul | -2.7 | -1.0 | +0.9 | 33% |
| Aug | -2.7 | -0.8 | +0.2 | 35% |
| Sep | -0.6 | -0.3 | -0.4 | 55% |
| Oct | +2.3 | -1.3 | -2.1 | 28% |
| Nov | +1.1 | +0.6 | +0.1 | 65% |
| Dec | -2.1 | +2.8 | +1.8 | 80% |
Source: EIA Weekly Petroleum Status Report historical data, Table 1 stocks (psw01.xls: commercial crude excluding SPR, total gasoline, distillate fuel oil), weeks ending January 2, 2015 to December 27, 2024, downloaded and computed by PetroEyes on October 29, 2026.
Gasoline: the winter build is real, the summer draw is not the big one, and autumn is the low point.January is the strongest month on the chart, a build of 3.9 million barrels in the average week, and 89 percent of January weeks in the decade were builds. December adds 2.8 million a week. Stocks then peaked in January or February in nine of the ten years; the exception was 2020, when the peak came on April 17. The hardest draws are in March, 2.9 million barrels a week on average, with only 11 percent of March weeks building. July and August draw about 1.0 and 0.8 million a week, and June is slightly positive. The autumn "rebuild" does not exist in this data: October averages a 1.3 million barrel draw, and the year's lowest gasoline level fell in November in seven of the ten years, in August, September or October in the other three.
Distillate: half right. Distillate does build in summer, modestly, between 0.2 and 0.9 million barrels a week from May to August. But its sharpest month is not deep winter. October averages a 2.1 million barrel draw, and only 12 percent of October weeks built. The annual low came in October or November in six of the ten years, and the annual high came between December 29 and February 3 in eight of them, after December builds that average 1.8 million a week. February and March then draw about 1.4 and 1.3 million. January is the month to read carefully: its average is a 1.3 million build, but the median week drew 0.3 million, and just under half of January weeks built. A few very large builds lift the mean, so a January distillate draw is not the anomaly the average suggests.
Crude: builds in late winter, draws in summer, and two odd months.Commercial crude stocks build from January to April, most strongly in February (3.6 million a week, 82 percent of weeks building), and draw from May through September, by about 2.3 to 2.7 million a week in June, July and August. Two months break the simple summer-winter story. October builds 2.3 million a week on average, and December draws 2.1 million, with only 24 percent of December weeks building. The year's low point for crude fell in September in five of the ten years and at the turn of the year in three. The data record that pattern, not its causes, so read the October and December rows as facts about the calendar to expect, and check the refinery block of the week for the explanation. Our refinery turnarounds guide covers the maintenance season that cuts crude runs in spring and fall.
Here is how the table changes a reading. In the report for the week ending October 23, 2026, total gasoline stocks rose 3.870 million barrels, distillate fell 1.462 million and commercial crude fell 1.892 million. Against zero, that reads as one build and two draws. Against the October averages above, it reads differently. Gasoline came in about 5.2 million barrels above the October norm of a 1.3 million draw, in a month when barely more than a quarter of weeks build at all. Distillate drew, as 88 percent of October distillate weeks did in the decade, but by about 0.6 million barrels less than the October average. And crude drew in a month whose average week builds 2.3 million, a gap of about 4.2 million barrels. Read against zero, the crude line looks like a small draw; read against the calendar, it is the largest seasonal surprise of the three. One week proves nothing about a trend, and the four-week rule from section 7 still applies, but the calendar is the right yardstick for the first read.
These averages come with limits. Ten years is a short sample, and it includes 2020, so treat a monthly average as a rough expectation rather than a forecast. Grouping by month blurs timing within the month: the March gasoline draw and the October crude build are concentrated in some weeks more than others. And the EIA's own comparison in the report is a five-year average for the same week of the year, which is tighter than a ten-year monthly mean and moves as old years roll out. Use the EIA's five-year line for the official comparison, and this calendar to know which months make a draw or a build ordinary.
9. Your Wednesday checklist
Print this sequence and run it in order. It takes about ten minutes once the report is familiar, and it is deliberately ordered so that the hardest data speaks first.
- Stocks versus expectation. Note the crude, gasoline, and distillate changes against the surveyed expectation, not against zero. Record the surprise, in barrels, for each.
- Cushing split. Check whether the crude move happened at Cushing or elsewhere. A national draw with a Cushing build is a different benchmark signal than the reverse.
- Runs and utilization. Did refinery inputs confirm the product moves? A gasoline draw alongside falling runs is a supply event. A draw alongside rising runs is a demand event.
- Demand on the average. Compare gasoline and distillate supplied on the four-week average. Ignore the single-week swing unless the average turns with it.
- Days of supply. Run the three divisions from section 5 and place each result against its recent range. Tightness is a level, not a change.
- Trade lump check. If imports or exports explain the crude move, check the four-week average before believing it. Tanker timing fakes more weekly crude stories than any other line.
- Price confirmation. Compare the physical reading with the spot prices in the report and the live benchmarks on our WTI price page. When price and physics disagree, write down which one you trust and why. That note is your audit trail next Wednesday.
10. Limits to keep taped to the monitor
- Weekly production is an estimate. The monthly Petroleum Supply Monthly, published with a longer lag, is the benchmark. The weekly figure is not usually revised after the fact; instead the EIA may re-benchmark the model in a Short-Term Energy Outlook week, so a sudden step in weekly production can be a model reset rather than a change in the field.
- Product supplied is a proxy. It is disappearance from primary storage, not metered end-use. It is steadier over months than over weeks, and even a month of weekly figures can land a few percent away from the monthly survey, as the product supplied guide measures.
- National totals hide districts. PADD-level tightness drives regional prices, including what drivers pay. Our gas prices page and the rockets and feathers deep dive connect those regional physical signals to the pump.
- One week is not a balance. The annual balance emerges from hundreds of weekly prints. The report rewards readers who keep a running table - which is why our inventory terminal archives each week instead of overwriting it.
- Rig counts answer a different question. The weekly Baker Hughes count, tracked on our rig counts page, leads production by months, not days. Do not use it to explain this week stock change.
11. Frequently asked questions
When is the EIA weekly report released?
Normally Wednesday at 10:30 a.m. Eastern, covering the week ended the prior Friday. Some federal holiday weeks, mostly those with a Monday holiday, move it to Thursday. The companion natural gas storage report follows on Thursday - our guide to reading the EIA Weekly Natural Gas Storage Report covers that release line by line, and we walk through its benchmark on the natural gas price page.
What is the difference between a build and a draw?
A build is a week-on-week increase in a stock level; a draw is a decrease. The terms describe direction only. Whether either matters depends on the seasonal expectation, the four-week trend, and which flows caused it, as sections 7 and 8 lay out.
Why did oil fall on a draw, or rise on a build?
Because the market had priced a different number. A small draw after a surveyed expectation of a large draw is, in trading terms, a relative build. Regional splits, especially Cushing, and the forward curve reaction can also overpower the headline within the same hour.
Which single number matters most?
There is not one, and distrust anyone who names one without conditions. If forced to rank, crude stocks against expectation set the headline, refinery runs explain it, and days of supply judge it. The checklist in section 9 is built on that ranking.
Is product supplied the same as demand?
It is the EIA weekly proxy for demand: volume leaving primary storage. It is not a direct measurement of fuel burned. Use the four-week average for trend claims, and expect the monthly report to refine the picture.
Where can I see the numbers each week without a paid service?
The EIA publishes the full report free on its website, and PetroEyes archives the core weekly series in the EIA inventory terminal, with benchmark prices on the WTI price page and published outlooks collected on the oil price forecast hub.
12. Related reading
- Strategic Petroleum Reserve Levels - the weekly reserve level since 1982 and the 2026 exchanges, with cover in days.
- US Refinery Capacity in 2026 - why refinery closures change the denominator behind the weekly utilization percentage.
- How to Read the EIA Short-Term Energy Outlook - how the weekly data feeds the EIA's monthly forecast.
- EIA Crude Oil Inventory and Weekly Petroleum Status Report - the archived weekly data behind this guide.
- EIA Report Schedule, Explained - when this report and the week's other releases actually come out, including holiday shifts.
- API vs EIA Inventory Reports, Explained - why the Tuesday API preview and this Wednesday report tally the same week differently.
- Product Supplied, Explained - the demand lines of this report: what the disappearance measure counts, and what it misses.
- WTI vs Brent Spread, Explained - why the Cushing line in this report moves the benchmark spread.
- Refinery Crack Spreads, Explained - the margin that decides how hard refineries run each week.
- Natural Gas (Henry Hub) Price Today - the Thursday storage companion to the Wednesday petroleum report.
- Rig Counts - the slower supply signal that leads weekly production by months.
- How to Read the CFTC Commitments of Traders - the Friday positioning report that follows the Wednesday inventory data.
Sources checked October 29, 2026: EIA, Weekly Petroleum Status Report explanatory notes (Appendix B) (holiday releases delayed by one day; production modelled from the Short-Term Energy Outlook and re-benchmarked in outlook weeks; imports from the weekly importer survey with a smoothed ratio; exports from Customs and Border Protection ACE filings; revision policy); EIA, WPSR historical Table 1 data (psw01.xls) (weekly stocks used for the section 8 calendar and the week ending October 23, 2026 changes); EIA, WPSR release schedule and WPSR archive (2026 exceptions; Thanksgiving-week releases 2015 to 2025). Monthly averages, peak and low dates and the October comparisons are PetroEyes arithmetic on those files.
Scope note: Weekly petroleum figures are from the EIA Weekly Petroleum Status Report as archived in PetroEyes market data, with the week-ending date shown for each series; weekly EIA flow and production figures are preliminary estimates subject to revision. The days-of-supply calculations are hypothetical cover arithmetic using those published levels and rates, labelled as such, and do not predict prices.