US Natural Gas Storage Before Winter 2026-27

Working gas in storage in the Lower 48 stood at 3,801 billion cubic feet (Bcf) for the week ending Friday, October 30, 2026, the last EIA Weekly Natural Gas Storage Report reading before this page was written. The week added 77 Bcf, the fifth straight weekly build of 67 Bcf or more, and it closed an injection season that lifted storage from a March 20 low of 1,818 Bcf by 1,983 Bcf over 32 weeks. That is the headline. The rest of the story is in the split: the East and Midwest regions ended October within 2 to 3 percent of their five-year maximums, while the South Central region, the largest of the five, sits 137 Bcf below its maximum and 88 Bcf below a year ago.

Storage matters because it is the only buffer between winter demand and daily production. The United States produces gas every day and burns it every day, but in winter the burn runs far ahead of production for months, and the difference comes out of storage. A storage report therefore answers one question each week: how big is the buffer, and how fast is it being filled or emptied. This page answers it from the PetroEyes archive of the EIA report, which holds the Lower 48 total and the five regional series week by week for the past 52 weeks, from November 7, 2025 through October 30, 2026. Every figure below comes from that archive or from the PetroEyes market data file, and every derived number is this page’s arithmetic on those two files.

Two cautions before the tables. First, the five-year figures here are the five-year range (minimum, maximum, average) stored with each series in the archive, not the EIA’s same-week five-year average that the full release prints next to the current level. The archive’s average is a range average across the whole five-year window, so it runs lower than a same-week average in October, when storage is seasonally near its peak. Treat the comparison as a position inside the range, not as the EIA’s surplus-or-deficit test. Second, nothing here forecasts a price. Storage is a stock, and the price reacts to the weekly flow against weather, which this page does not predict.

The headline: 3,801 Bcf, and where it sits

The October 30 level of 3,801 Bcf is 104 Bcf below the same week a year earlier, when the archive puts the year-ago level at 3,905 Bcf, a deficit of 2.7 percent. It is also 149 Bcf below the archive’s first reading, 3,950 Bcf on November 7, 2025, a week further into last November. Set against the five-year range, the picture flips: 3,801 Bcf is 95.7 percent of the five-year maximum of 3,972 Bcf and 133.8 percent of the five-year range average of 2,840 Bcf. The two comparisons disagree because they measure different things. The year-over-year test says the buffer is a little thinner than last year. The range test says the buffer is near the top of anything the past five years produced. Both are true, and the winter will be decided by the draw, not by either starting point alone.

The last column is the one to keep in mind while reading the rest of the page. The South Central region holds 31.8 percent of all working gas, the Midwest 29.2 percent and the East 24.4 percent, so those three regions together hold 85.4 percent of the national stock. The Mountain and Pacific regions hold 6.8 and 7.8 percent. A national total can therefore move on the back of one region, and a regional problem can hide inside a comfortable national number. The regional section below takes them in order of size.

One more scale check: the capacity page on this site puts US working gas capacity at 4.83 trillion cubic feet across 408 fields in 31 states. Against that capacity, 3,801 Bcf is 78.7 percent full. Capacity is not a target, and no field is ever filled to its nameplate, but the ratio puts the 95.7 percent of the five-year maximum in context. Storage is near the top of what recent years have held, and about four-fifths of what the fields can physically hold.

The injection season, month by month

The season ran from the March 20 low of 1,818 Bcf to 3,801 Bcf on October 30, a gain of 1,983 Bcf over 32 weeks at an average of 62.0 Bcf a week. The pace was uneven. June delivered the biggest single weekly injection in the archive’s window, 108 Bcf in the week ending June 5, and the slowest week, 15 Bcf in the week ending August 21, came in the hottest part of summer, when gas-fired power generation for air conditioning competes with storage injections for the same molecules. October then finished strong: the five weekly readings of the month added 85, 67, 77, 80 and 77 Bcf, and the last eight weeks of the season added 547 Bcf, more than a quarter of the whole season’s build.

Read the table as two seasons, not twelve months. November 2025 through March 2026 is the withdrawal season: storage fell from 3,950 Bcf to 1,851 Bcf, a draw of 2,099 Bcf across the five month-end readings, with January alone taking 910 Bcf out. April through October is the injection season: storage rose from 1,851 Bcf to 3,801 Bcf. The turn came in March, when the month ended almost flat, down 25 Bcf, after the low of 1,818 Bcf on March 20. April added 291 Bcf, May 436 Bcf, and by the end of June the stock had passed 2,900 Bcf. July and August were the slow months, adding 195 and 97 Bcf, which is the usual summer pattern: injections compete with power demand. September picked up again at 201 Bcf, and October, at 386 Bcf, was the second-strongest month of the injection season, behind May’s 436 Bcf.

October’s 386 Bcf was the second-biggest monthly build of the season, behind May’s 436 Bcf, and it came from five straight builds of 67 Bcf or more in weeks when the heating season would usually have started to nibble at the stock. Mild autumn weather is the usual explanation for late injections, and the report itself does not give a reason; it gives the flows. What the flows say is that the injection season ran a week or two longer than the calendar suggests, and the stock that entered November was larger than the pace of September implied.

The regions, one by one

The East region ended October at 926 Bcf, up 13 Bcf on the week and 17 Bcf above a year ago. Its five-year range is 229 to 943 Bcf, so 926 Bcf is 98.2 percent of the maximum the region has held in five years, the fullest reading of any region against its own range. That matters more than the national total, because the East serves the Northeast heating market, where winter demand is concentrated and where pipeline capacity into New England is the tightest in the country. A full East region does not guarantee low prices in Boston in January, since a cold snap draws the region down fast and the pipes can only bring so much in per day, but it takes the shortage scenario off the table at the start of the season. The East is the region to watch first in every Thursday release from here on.

The Midwest, the second-largest region at 1,111 Bcf, is in nearly the same shape: up 23 Bcf on the week, 6 Bcf above a year ago, and 97.4 percent of its five-year maximum of 1,141 Bcf. The Midwest is the storage workhorse of the system, a dense network of depleted fields in Michigan, Illinois and Ohio that sits next to both production and demand, and it holds 29.2 percent of the national stock. A Midwest reading near its maximum means the fields that supply Chicago and the industrial Midwest enter winter with their cushion intact. The region’s five-year range average is 743 Bcf, so the current level sits far above the middle of its own history, as the national total does.

The South Central region is the exception, and it is the biggest one. At 1,208 Bcf it is still the largest regional stock, 31.8 percent of the total, and it added the most gas in the final week, 40 Bcf of the national 77 Bcf. But it is 88 Bcf below a year ago, the only region with a year-over-year deficit larger than 25 Bcf, and it sits at 89.8 percent of its five-year maximum of 1,345 Bcf, the lowest percentage of any region. The gap between the South Central level and its maximum is 137 Bcf, more than half the entire Mountain region’s stock of 260 Bcf. South Central storage is dominated by salt caverns along the Gulf Coast, which inject and withdraw faster than the depleted fields that dominate the East and Midwest, so the region is built to swing. A thinner South Central stock means less fast-response supply for Gulf Coast industry and LNG export plants, and it is the one regional number that keeps the national total from reading as full as the East and Midwest suggest.

The two small regions complete the picture. The Mountain region holds 260 Bcf, up 2 Bcf on the week, 17 Bcf below a year ago and 88.7 percent of its five-year maximum of 293 Bcf. The Pacific region holds 296 Bcf, down 1 Bcf on the week, the only regional draw in the final report, 22 Bcf below a year ago and 91.9 percent of its five-year maximum of 322 Bcf. Together they hold 556 Bcf, less than the East alone, and their year-over-year deficits of 17 and 22 Bcf are small in absolute terms. The Pacific draw is the detail to note: a region drawing gas in the last week of October, while the other four injected, is an early sign that its withdrawal season has already started, which fits a region whose storage serves a milder climate and whose injections end earlier.

Last winter’s draw as the yardstick

The most useful comparison for a storage stock is not another stock but a draw. The archive holds a full withdrawal season, from the first reading of 3,950 Bcf on November 7, 2025 to the low of 1,818 Bcf on March 20, 2026, and that season is the yardstick for the winter ahead. The draw was 2,132 Bcf over 19 weeks, an average of 112.2 Bcf a week, with 18 of the 19 weeks showing a net withdrawal. The single biggest week took 359 Bcf out, in the week ending January 30, 2026. January was the heavy month, with 910 Bcf withdrawn between the December 26 and January 30 readings, and February took another 578 Bcf.

Apply last winter’s draw to this winter’s starting stock and the arithmetic is straightforward. A repeat of the 2,132 Bcf draw would take storage from 3,801 Bcf to 1,669 Bcf by the end of the season, 149 Bcf below where last winter ended, because this winter starts 149 Bcf below the archive’s November 2025 reading. At last winter’s average weekly draw of 112.2 Bcf, the 3,801 Bcf stock would last 33.9 weeks, far longer than a withdrawal season of about 20 weeks. Those two sentences are arithmetic, not a forecast, and they point in opposite directions on purpose. The first says the cushion is thinner than last year by exactly the year-over-year gap. The second says the cushion is far larger than a single winter’s draw. Both are true because a winter draws roughly half the stock, so a 149 Bcf gap is small against a 2,132 Bcf draw and large against nothing at all.

What would change the arithmetic is the weather, which is also what the storage number cannot tell you. The January 2026 week that withdrew 359 Bcf is the warning: a single cold week can take out more than three times the average weekly draw. Three such weeks in a row would take over 1,000 Bcf, half the injection season’s build, in less than a month. That is why traders watch the weekly change against the expected change for the season rather than the level alone, and why the five-year comparison in the EIA release is a same-week average rather than a range. A level that looks comfortable in October can look thin by February if the draws run hot, and a level that looks thin in October can last the winter if the draws run mild. The storage report is a weekly update to that bet, not a verdict on it.

What the price did

Henry Hub, the US gas benchmark, closed at $3.20 per million Btu on Monday, November 9, 2026, in the PetroEyes market data file. That is up $0.31, or 10.7 percent, from $2.89 on November 3, and up from $3.00 on November 6. Over the 52 trading sessions in the file, Henry Hub ranged from $2.60 on October 26 to $3.30 on September 24, so the November 9 close sits near the top of the recent range. The price rose in the same week storage printed near its five-year maximum, which is a reminder that the two series answer different questions. Storage reports the stock. The price reports the market’s view of the flow: expected withdrawals, expected production, LNG export demand, and the weather forecast for the next two weeks.

The wider energy market moved the other way in the same sessions. WTI crude closed at $86.43 a barrel on November 9, down from $96.09 on November 3, a drop of 10.1 percent in four sessions, and Brent fell with it. Gas rising while oil falls is not unusual; the two markets share some drivers and not others, and a gas price move in November is more often a weather and storage story than an oil story. The gas bill guide on this site follows the hub price to the household bill, and the lag matters: the price a household pays this winter was largely set by the gas its utility bought months ago, so a November move at Henry Hub reaches bills slowly, if at all.

How to read the next release

The next storage report lands Thursday, November 12, 2026, at 10:30 a.m. Eastern, covering working gas as of Friday, November 6. It is the first release of the withdrawal season, and the first question it answers is whether the season has actually turned: a net withdrawal, however small, means injections are over, while another injection would mean the mild autumn is still adding to the stock. The second question is the size of the change against the five-year average change for the same week, which the full EIA release prints and which the archive’s range figures do not replace. A draw larger than the seasonal norm tightens the winter even if the level stays high; a build in November loosens it further.

The third question is the regional split, and it is the one this page’s tables are built to support. Watch the East first: a draw there in November, while the national total still builds, would mean the region nearest winter demand is already turning. Watch the South Central region second: it ended October 88 Bcf below a year ago, and a weak start to its withdrawal season would confirm that the largest region is the thinnest. The storage report guide on this site walks through the release line by line, including the salt-cavern detail and the revision note that can restate a prior week without any gas moving. Read the new release against Table 1 here, region by region, and the story will usually be visible in the first minute.

What this page leaves out

The archive behind this page holds 52 weeks, from November 7, 2025 to October 30, 2026. That is one withdrawal season and one injection season, which is enough for the yardstick in section 4 and not enough for a longer history. The page does not compare this winter’s starting stock with the winters before 2025, and it does not use the EIA’s same-week five-year average, because the archive stores a five-year range average instead. Anyone who needs the same-week comparison should read the full EIA release, which prints the current level next to the year-ago level and the five-year average for the same week.

The page also says nothing about production, LNG exports, or pipeline capacity, all of which decide how fast storage is drawn. Production sets how much gas is available to burn before storage is touched; exports set how much leaves the country before it can be stored; pipelines set how fast gas can move from where it is stored to where it is burned. The production-by-state guide, the LNG terminals guide and the Henry Hub benchmark guide on this site cover those three, and the winter fuels outlook guide covers the EIA’s own seasonal forecast, which this page deliberately does not repeat. Storage is one leg of the winter balance. It is the leg with the cleanest weekly number, which is why it gets its own page.

Data and sources: U.S. Energy Information Administration, Weekly Natural Gas Storage Report, week ending October 30, 2026 (released November 5, 2026), as archived in the PetroEyes inventory archive, checked November 10, 2026. PetroEyes market data archive for Henry Hub and WTI closes (front-month futures, November 9, 2026), checked November 10, 2026. EIA report timing from the storage report guide on this site. Working gas capacity of 4.83 trillion cubic feet from the storage capacity guide on this site. Regional levels, weekly changes, year-over-year changes, five-year range figures, monthly changes, draw arithmetic, percentages and the three winter-ending scenarios are this page’s arithmetic on those files. The five-year range figures are the archive’s stored five-year range, not the EIA’s same-week five-year average.