How to Read the EIA Weekly Natural Gas Storage Report

Every Thursday at 10:30 a.m. Eastern, the US Energy Information Administration publishes the Weekly Natural Gas Storage Report. It is the fastest official count of how much gas sits in underground storage across the Lower 48 states, broken into five regions, with a comparison to the same week a year ago and to the five-year average. Gas traders treat the release the way oil traders treat the Wednesday petroleum report: the headline moves prices in seconds, and the detail decides whether that move holds. A single number - the net change in working gas - gets quoted as bullish or bearish before most readers have seen which region drove it, whether salt storage moved with it, or how far the print landed from what the market expected.

This guide walks through the report line by line, in the order a careful reader works through it. Section 1 defines what the report counts and what it leaves out. Section 2 covers the calendar and the survey behind the numbers. Section 3 maps the five regions and the salt split inside South Central. Section 4 lays out the latest week from PetroEyes archived data for the week ending October 2, 2026, computed at build time so the math can be checked against the source file. Later sections cover the injection and withdrawal seasons, the five-year average test, salt deliverability, and a Thursday checklist. The companion guide to the EIA Weekly Petroleum Status Report covers the Wednesday release in the same style, and our natural gas price page tracks the Henry Hub benchmark that prices the result.

1. What the report counts: working gas only

The report counts working gas in underground storage. Working gas is the volume above base gas that operators can withdraw and sell. Base gas, also called cushion gas, remains in the reservoir to hold pressure and keep the facility able to deliver gas at a useful rate through the withdrawal season. It is not counted in the weekly total, and it is not available to meet a cold snap next week. The distinction matters for scale. A facility that holds 100 billion cubic feet (Bcf) of total gas might hold 50 Bcf of working gas and 50 Bcf of base gas, and only the 50 Bcf moves in the weekly report.

Base gas takes a different share of capacity by facility type. The EIA notes that in the Lower 48, base gas is about 30 percent of capacity in salt caverns and about 50 percent in nonsalt reservoirs, which are depleted oil and gas fields and aquifers. Salt caverns can sometimes pull briefly from the volumes normally held as base gas because brine or liquids can be moved to manage pressure, which gives salt storage operating room that a depleted field does not have in the same way. The weekly report does not publish base gas. It publishes the working gas level, the net change from the prior week, and the implied flow that the change represents.

Storage facilities themselves fall into three kinds, and the kind shapes how fast the gas can move. Depleted reservoirs are former oil or gas fields turned into storage; they hold large volumes and are the backbone of seasonal storage in the East and Midwest. Aquifers are water-bearing rock formations converted to storage, concentrated in the Midwest, and they also move gas slowly. Salt caverns are solution-mined cavities in salt formations along the Gulf Coast; they hold less gas per site but can inject and withdraw at high rates and can complete more than one fill-and-empty cycle in a single year. Section 7 returns to that rate difference, because it explains why the South Central salt line often moves first when weather turns.

Two things the report does not do deserve a clear statement. It does not survey gas in pipelines, gas in liquefied form at export terminals, or gas held above ground. It also does not measure gas burned that week. The net change is an implied flow: this week working gas level minus last week working gas level, with revisions and reclassifications footnoted when they occur. Demand, supply, and trade all sit behind that one subtraction, and the reader has to infer them from weather, production, and flow data published elsewhere. The report is a stock count with a change attached, not a full balance sheet.

2. The Thursday calendar and the EIA-912 survey

The standard release is Thursday at 10:30 a.m. Eastern. The EIA posts a holiday schedule with the exceptions: in weeks with a federal holiday, the release moves to Wednesday at noon Eastern or to Friday at 10:30 a.m. Eastern, depending on which day the holiday falls. The report describes working gas as of 9 a.m. Eastern on the Friday of the report week, so the publication lag is six days. A cold snap that starts on Saturday does not appear until the report covering the following Friday, published nearly two weeks after the weather began. Gas futures trade the forecast in real time; the storage report confirms what already happened underground.

The data come from Form EIA-912, the Weekly Underground Natural Gas Storage Report. Operators in a sample of underground storage companies report their working gas by region, as of that Friday 9 a.m. mark, with reports due after the weekend. The EIA combines the sample with estimates for operators outside the sample to produce regional and national estimates. The agency describes the Weekly Natural Gas Storage Report as the Department of Energy only designated Principal Federal Economic Indicator, which is a formal way of saying the release is produced under strict timing and revision discipline, and the market treats it that way.

Revisions are part of the design, not a failure of it. Operators submit revised figures when a prior week working gas measure changes by 500 million cubic feet or more in a region, and the EIA carries reclassification notes when gas moves between the working and base categories after an engineering or accounting review. In practice, most weeks print without a revision, and most revisions are small against a national total near 3,500 Bcf. The reading habit that follows is simple: trust the level, respect the change, and check the footnotes before treating a one-week surprise as a new trend. A revised prior week can turn a reported withdrawal into a smaller one without any gas moving at all.

The Wednesday petroleum report and the Thursday gas report make a pair. The petroleum report, covered in our guide to the Weekly Petroleum Status Report, prints crude and product stocks, refinery runs, and product supplied. The gas report prints the storage side of the other main US fuel. Reading them on consecutive days gives a complete weekly picture of US hydrocarbon inventories before the Friday rig count adds the drilling side. None of the three releases replaces the others, and each one answers a narrower question than headlines suggest.

3. The five regions, plus salt inside South Central

The report breaks the Lower 48 into five regions: East, Midwest, South Central, Mountain, and Pacific. Alaska and Hawaii sit outside the weekly regional table. The EIA also splits South Central into salt and nonsalt, because the two kinds of storage in that region behave differently enough to change the national story. The total United States line in the weekly table is the Lower 48 total; the five regions sum to that total subject to rounding, which the EIA notes as totals may not equal sum of components because of independent rounding.

East. Storage in the Northeast and Southeast, weighted toward depleted fields that serve winter heating demand in the most densely populated gas market in the country. The East often leads the withdrawal season story because its storage sits closest to the demand it covers.

Midwest. Storage across the Great Lakes and upper Midwest, a mix of depleted fields and aquifers. The Midwest holds large seasonal volumes and is read as the second winter buffer after the East.

South Central. Storage in the producing region - Alabama, Arkansas, Kansas, Louisiana, Mississippi, Oklahoma, and Texas in the current regional definition. This is the only region split into salt and nonsalt. Salt caverns cluster on the Gulf Coast near production, pipeline hubs, and liquefied natural gas (LNG) terminals. Nonsalt storage in South Central behaves like storage in the East and Midwest: larger, slower, seasonal.

Mountain and Pacific. Smaller regional totals in the West. Mountain storage serves Rocky Mountain and Southwest demand with a mix that includes high-deliverability sites. Pacific storage, much of it in California, is watched closely during west coast supply events even though its national share is modest, because pipeline links between the Pacific and the rest of the system are thinner than links inside the East and Midwest.

Geography explains why a national total can mislead. A 60 Bcf national injection with 50 Bcf landing in South Central salt and 10 Bcf spread across the other four regions describes gas parked near Gulf Coast supply and LNG demand. The same 60 Bcf with 40 Bcf landing in the East and Midwest describes gas placed where winter heating will need it. The headline is identical. The winter meaning is not. Reading the regional lines before the total is the cheapest accuracy upgrade available to a Thursday reader, and section 4 shows the split in the latest archived week.

4. The latest week, laid out in full

The table below is the working asset for this guide. It is computed directly from PetroEyes archived inventory data for the week ending October 2, 2026, the most recent week in the file. Columns show the current working gas level, the week-on-week net change (the implied flow), the change against the same week a year ago as archived in the file, the five-year average level for the region, and the surplus or deficit against that average in both Bcf and percent. The five-year figures are the archive five-year range averages stored with each series; they let a single print be judged against its seasonal reference without opening a second document.

Read the current print through the table rather than through the headline. For the week ending October 2, 2026, Lower 48 working gas stood at 3,500 Bcf after a net change of +85 Bcf on the week, leaving the total +664 Bcf (+23.4%) against the five-year average of 2,836 Bcf in the archive. That is an injection - a net gain - in a week that falls inside the traditional refill season, and its size only becomes informative next to the five-year average change for the same week, which the full EIA release prints and the archive table frames through the level comparison. The regional split then shows where the gas landed: the East and Midwest added the steady seasonal volumes their winter role requires, South Central contributed the largest single regional share as it often does near the Gulf Coast supply base, and the Mountain and Pacific regions moved by small amounts that rarely drive the national story on their own. One weekly build, read this way, is a statement about refill pace and placement, not a verdict on winter prices.

The year-ago column does different work. It strips out the seasonal shape and asks whether storage discipline this year matches last year at the same point in the calendar. A total that sits below the year-ago level after a summer of strong power demand and steady LNG feedgas tells a tightening story even if the latest week injected. A total that sits above the year-ago level after a mild winter tells the opposite story even if the latest week withdrew. The five-year comparison in the last two columns then judges both, because one prior year can itself be the outlier. Professional readers quote the five-year surplus or deficit first, the year-ago gap second, and the weekly change third. Headlines quote them in the reverse order.

5. Injection season, withdrawal season, and shoulder months

US gas storage follows a calendar set by weather. The traditional refill or injection season runs from April through October: production that is not burned for power or heating, plus pipeline imports, exceeds demand, and the surplus goes underground. The withdrawal season runs from November through March: heating demand exceeds current supply, and storage makes up the difference. The market shorthand calls the end of October the start of winter and the end of March the start of refill, and end-of-season levels get quoted against those dates all year.

Summer has complicated the clean two-season picture. Gas-fired power generation for air conditioning now pulls heavily on supply in July and August, so weekly injections shrink in heat waves and can flip to small withdrawals in extreme weeks even in the middle of refill season. A soft injection in August therefore does not carry the same meaning as a soft injection in May. In May it signals supply or pipeline trouble. In August it often just records power burn. The five-year average for the specific week already contains that summer effect, which is another reason the comparison in section 4 anchors on the average rather than on the sign of the change.

Shoulder months - roughly April and October, with March and November as transition edges - flip between injections and withdrawals on modest weather swings. A late cold snap in April can produce a withdrawal in refill season; a warm November can produce an injection in withdrawal season. Neither event breaks the seasonal frame. The error to avoid in shoulder months is annualizing a single flipped week. One withdrawal in April does not delay refill in any mechanical sense; it lowers the starting level from which refill proceeds, and the following weeks of injections either recover the pace or they do not. The trend across four weeks, not the sign of one week, carries the information - the same four-week discipline our petroleum report guide applies to crude flows.

End-of-season levels matter because they set the starting buffer for the next season. A winter that ends with storage near the bottom of the five-year range leaves refill season with ground to recover before the market can relax about the following winter. A refill season that ends with storage near the top of the range in late October leaves winter with a cushion that mutes the price effect of an ordinary cold snap and concentrates risk in an extended one. The gas futures curve prices that cushion in advance, which is why the March-April spread - the gap between the last winter contract and the first spring contract - reacts so sharply to late-winter storage prints. Our contango versus backwardation guide walks through that spread and why traders watch it as a storage verdict.

6. The five-year average test and the expectation gap

Markets trade the gap between the print and the expectation, not the print alone. Before the Thursday release, analyst surveys circulate an expected net change, built from weather during the report week, production estimates, LNG feedgas, and pipeline flow data. A 75 Bcf injection against an 85 Bcf expected injection is a tight print - less gas went underground than the market had priced - and prices often rise on it even though storage grew. A 95 Bcf injection against the same 85 Bcf expectation is a loose print and prices often fall. The sign of the change settles nothing; the sign of the surprise settles the first move.

The five-year average test is the slower, more durable version of the same idea. The EIA prints the current level, the year-ago level, and the five-year average level for the report week, plus the percent difference from each. A surplus to the five-year average means more gas is stored than in a typical recent year at this point in the calendar; a deficit means less. Percent matters alongside Bcf: a 100 Bcf deficit in late October, against a five-year average near 3,600 Bcf, is a different risk than a 100 Bcf deficit in late March, against an average roughly half that size, because the denominator changes with the season. Quote both numbers together or quote neither.

Three judgements follow from the test, in order. First, level versus average judges the buffer: surplus means cushion, deficit means exposure. Second, trajectory versus average judges the pace: a deficit that shrinks by 10 Bcf a week during refill is closing, and a surplus that shrinks by the same amount during winter is being spent. Third, weeks remaining judge whether pace can finish the job: refill has a fixed number of weeks before November, and a deficit that is closing too slowly to reach the average by late October is information even if every individual week injected. That third judgement - required pace versus actual pace over the weeks left - is the arithmetic behind most end-of-season storage forecasts, and it needs nothing beyond the weekly levels and a calendar.

Regional surpluses refine the national test. A national surplus built mostly in South Central, with the East sitting at a deficit to its five-year average in October, leaves the demand centre least covered where winter will call first. Pipeline capacity between regions is finite, so gas stored in the wrong region helps less than gas stored in the right one. The table in section 4 is built for exactly that check: read the Lower 48 surplus, then scan the East and Midwest rows before repeating it.

7. Salt versus nonsalt: deliverability is the hidden variable

Two storage systems share the national total. Nonsalt storage - depleted fields and aquifers - holds the bulk of working gas and moves it at moderate rates, filling through summer and emptying through winter in one long cycle. Salt storage holds less gas, concentrated in South Central caverns, and moves it fast in both directions. Salt caverns can be cycled several times a year: filled in a mild stretch, emptied into a cold snap, and refilled when prices soften, all inside one season. That flexibility makes salt the swing storage of the US system.

Deliverability is the rate at which a facility can withdraw gas on a given day, and it falls as a nonsalt field empties, because reservoir pressure drops with the level. A depleted field at 30 percent of working capacity cannot deliver at its nameplate rate. Salt caverns hold deliverability far better as they empty, because pressure is managed with brine and the cavern geometry differs. The practical consequence lands in late winter: after months of withdrawals, the nonsalt system delivers more slowly exactly when a late cold snap needs gas fastest, and salt storage carries a larger share of the daily response. A storage total that looks adequate in February can still be tight in daily deliverability terms if the remaining gas sits in low-pressure nonsalt fields far from the demand.

Watch the salt line for the first signal of stress or relief. In a genuine cold event, South Central salt posts the largest weekly withdrawal in the region table, because caverns respond first and pipelines pull from the most flexible source. In a warm spell during winter, salt posts the first small injection or the shallowest withdrawal, for the same reason. Nonsalt regions confirm the move over following weeks as the slower seasonal draw proceeds. Readers who check the salt split before the national change get the direction of the week right more often than readers who start from the total, and they get it from data the report prints every week.

LNG demand ties into this section at the Gulf Coast. Export terminals draw feedgas through the same pipeline network that serves South Central storage, so a terminal outage can leave gas stranded in the producing region and show up as a larger South Central injection or a smaller withdrawal in the next report, while a return to full feedgas does the reverse. Our LNG export dynamics guide explains the terminal side of that link, and our rig counts page tracks the drilling activity that sets future supply behind both.

8. Worked example: reading a hypothetical Thursday print

The example below is a labeled hypothetical - invented teaching numbers built to show the reading order, not a real release. Suppose a Thursday headline reads: US working gas rose 82 Bcf last week. The surveyed expectation was a 94 Bcf injection, the five-year average change for the week is a 92 Bcf injection, and the report week falls in late September, inside refill season. Here is the walk-through, in the order a careful reader runs it.

Step 1: score the surprise, not the sign. Storage rose, but by 12 Bcf less than expected and 10 Bcf less than the five-year average pace. The print is tight against both references. A headline that calls it bearish because injections are bearish has skipped the only comparison the market trades.

Step 2: place the level. Suppose the 82 Bcf build lifts the Lower 48 total to 3,430 Bcf, against a five-year average level of 3,360 Bcf for the week and a year-ago level of 3,390 Bcf. The level sits 70 Bcf (about 2.1 percent) above the five-year average and 40 Bcf above last year. Buffer: adequate. Trajectory: the surplus grew more slowly than the seasonal pace this week, which trims the cushion growth without removing the cushion.

Step 3: split the regions. In this hypothetical, the East injected 24 Bcf, the Midwest 27 Bcf, South Central 22 Bcf (salt up 3 Bcf, nonsalt up 19 Bcf), the Mountain 5 Bcf, and the Pacific 4 Bcf. The build landed where winter needs it - East plus Midwest took over three-fifths - and salt barely moved, which fits a normal refill week without a Gulf Coast disruption. Had salt taken 20 of the 22 South Central Bcf while the East lagged, the same national number would describe gas parked near supply rather than placed near demand.

Step 4: run the weeks-remaining check. Six refill weeks remain before November in the hypothetical. At the five-year average pace for those weeks, storage would add roughly another 280 Bcf and finish near 3,710 Bcf. At the slower pace of the latest week repeated, it would finish about 60 Bcf lower. Neither finish is a crisis against a five-year average end-of-October level; the check converts an anxious headline into a sized statement about the winter buffer.

Step 5: name what the report cannot say. The print does not reveal whether the shortfall came from production, power burn, LNG feedgas, or pipeline maintenance during the report week. Those causes live in daily flow and weather data, not in the stock count. The correct summary is one sentence: refill came in light against expectation and against the seasonal pace, storage still sits above the five-year average, and the build was placed in the consuming regions. Anyone who summarized the same print as a price call has added two steps the release does not contain.

9. Five common misreads

Misread 1: An injection is bearish and a withdrawal is bullish. Direction without a reference is noise. An injection below the expected and average pace tightens the balance; a withdrawal smaller than expected loosens it. Judge every change against the survey expectation and the five-year average change for the week before assigning a sign.

Misread 2: The national total describes the whole country. Regional placement decides winter usefulness. A surplus held in South Central does not cover an East deficit one-for-one because pipelines between regions are finite. Read the East and Midwest rows before quoting the Lower 48 row.

Misread 3: A surplus to the five-year average means prices must fall. The average is a buffer gauge, not a price formula. Weather forecasts, production trends, and LNG feedgas move prices between Thursday prints, and a surplus can coexist with rising prices when a cold forecast arrives. Storage judges how much shock the system can absorb; it does not schedule the shock.

Misread 4: Salt and nonsalt are interchangeable. They differ in rate, cycling, and pressure behaviour. Salt responds first in both directions and keeps deliverability as it empties; nonsalt holds the volume and slows as pressure falls. Late-winter adequacy is a deliverability question before it is a level question.

Misread 5: One week resets the season. A single print carries survey timing, weather timing, and revision risk. The EIA itself flags that totals may not equal the sum of components because of independent rounding, and operators revise prior weeks when a regional figure moves by 500 million cubic feet or more. Four weekly prints in the same direction, against the same seasonal reference, make a trend. One print makes a headline.

10. Your Thursday checklist

Work down the rows in order. Each row names where to look and what the check settles, so the headline cannot skip the queue.

Rows 1 through 6 come from the release itself and take minutes. Row 7 is the discipline: it forces the storage number to sit next to the price it is supposed to explain before the number becomes a claim. A Thursday take that skips rows 2 through 4 may still be a fine summary of the headline. It is not yet a storage reading.

11. Frequently asked questions

When is the EIA Weekly Natural Gas Storage Report released?

Thursdays at 10:30 a.m. Eastern, covering working gas in storage as of the prior Friday at 9 a.m. Eastern. The EIA posts a holiday schedule that moves the release to Wednesday at noon or Friday at 10:30 a.m. in weeks with a federal holiday. Our natural gas price page carries the Henry Hub benchmark that reacts to the print.

What is working gas, and how is it different from base gas?

Working gas is the gas in an underground facility that can be withdrawn for sale. Base gas, also called cushion gas, stays in the reservoir to keep pressure and deliverability. The weekly report counts working gas only. Base gas is a larger share of capacity in depleted fields and aquifers (about half) than in salt caverns (about 30 percent), which is one reason salt storage can cycle faster.

What do injection and withdrawal mean in the report?

An injection is a net gain in working gas during the report week - more gas went in than came out - and the report shows it as a positive net change and a positive implied flow. A withdrawal is a net loss, shown as a negative change. Injections dominate from spring through fall while withdrawals dominate in winter, with shoulder months flipping between the two.

Why does the market compare storage to a five-year average?

The absolute level means little without a seasonal reference. The EIA prints the current level next to the year-ago level and the five-year average for the same week. A surplus to the five-year average signals a loose balance for that week of the year; a deficit signals a tight balance. Traders judge the weekly change against the expected change for the season, not against zero.

What is the difference between salt and nonsalt storage?

Salt storage sits in salt caverns, concentrated in the South Central region. Nonsalt storage sits in depleted oil and gas fields and aquifers, spread across the East, Midwest, Mountain, and Pacific regions plus the nonsalt part of South Central. Salt caverns can inject and withdraw at higher rates and can be cycled more than once in a year. Nonsalt fields hold more gas but move it more slowly.

Where can I see the current storage number without a paid service?

The EIA publishes the full report free on its website each Thursday. PetroEyes archives the Lower 48 total and the five regional series from the same release on the analysis page, tracks the price response on the natural gas price page, and places the gas print next to the Wednesday petroleum report so the two weekly releases can be read together.

12. Related reading

• Natural Gas (Henry Hub) Price Today - the benchmark that prices each Thursday print, with the EIA weekly spot series and recent history.
• How to Read the EIA Weekly Petroleum Status Report - the Wednesday companion release: crude and product stocks, refinery runs, and days of supply.
• Contango vs Backwardation, Explained - how the gas futures curve, including the March-April spread, registers the storage position.
• LNG Export Dynamics - how Gulf Coast export terminals draw on the same supply pool that fills South Central storage.
• Baker Hughes Rig Count, Explained - the Friday drilling census behind future gas supply, and why rigs are not production.
• EIA Inventory Analysis - the archived weekly series behind the table in this guide, alongside crude and product stocks.