How Many People Work in Oil and Gas? US Jobs by Year

The Bureau of Labor Statistics counted 372,500 jobs, on 2025 annual average, in the three industries that find, drill and produce US oil and gas: 117,600 at oil and gas companies themselves, 42,700 at well drillers and 212,200 at oilfield service firms. That is 99,100 fewer than in 2019, a drop of 21%, even though US crude production rose 10.8% over the same six years to a record 13.66 million barrels a day. Add refineries and pipelines and the count reaches 492,100.

Industry groups quote much larger totals. The Texas Independent Producers and Royalty Owners Association (TIPRO) put US oil and gas direct employment at 2,043,859 for 2025, and “total direct and indirect jobs tied to the industry” at more than 19 million. Each is an honest count of a different thing, and the gap between 372,500 and 19 million is almost entirely a matter of where each source draws the industry’s boundary and whether it adds jobs at suppliers and in the wider economy. The definitions table further down sets each figure beside the rule that produces it.

US oil and gas payroll jobs, 2016 to 2026

Every number in Table 1 is a BLS annual average from the Current Employment Statistics (CES) survey, the monthly survey of employer payrolls behind the jobs figure in each first-Friday Employment Situation report. Annual averages exist only for the series that are not seasonally adjusted, so the table uses those throughout. The last row is a 2026 year-to-date average and is not directly comparable with full years, because oilfield employment has a seasonal pattern; the comparison with the same months of 2025 is in the section on 2026.

The upstream total (the sum of the first three columns) peaked at 471.6 thousand in 2019, slightly above 2018’s 470.4 thousand, and has not come close since; the post-pandemic high was 390.8 thousand in 2023. The producers themselves (NAICS 211) shrank by 52,200 jobs, or 30.7%, between 2016 and 2025, and almost all of that fell in two steps: 2016 to 2017 and 2019 to 2021. Refining and pipeline employment barely moved. Refineries went from 70,200 jobs in 2016 to 63,300 in 2025, and pipelines rose from 49,800 to 56,300, one of only two lines, with support activities, to finish the decade higher.

Support activities is the swing line. It added 64,200 jobs between 2016 and 2019, lost 74,700 in 2020 alone, regained 46,300 between 2021 and 2023, and has drifted down every year since. That pattern follows drilling and completion activity, which is what service companies sell. Our guide to reading the Baker Hughes rig count explains the weekly activity gauge that service-company hiring tends to trail.

What each BLS line counts, and who falls in which

The CES survey measures jobs, not people. BLS’s Handbook of Methods defines employment as “the number of persons on establishment payrolls” who received pay for any part of the pay period that includes the 12th of the month, and states that it is “the number of jobs, as opposed to the number of employees.” A person with two payroll jobs can appear twice. The survey excludes sole proprietors, the unincorporated self-employed and unpaid family workers, which matters in an industry with many independent landmen, consultants and owner-operators.

Each job lands in one industry, chosen by the main activity of the establishment that pays it, under the North American Industry Classification System (NAICS). The split that most often confuses readers is between NAICS 211 and NAICS 213. BLS describes oil and gas extraction as establishments that “operate and/or develop oil and gas field properties”, including exploration and “drilling, completing, and equipping wells.” Support activities for mining, NAICS 213, covers the same kinds of work when it is done “on a contract or fee basis.” The difference is who owns the well, not what the worker does.

Two consequences follow. When an operator lays off its own field staff and hires a service company to do the same work, jobs move from the 211 column to the 213112 column without any change in what happens at the well. And when a service company shuts a yard, the loss shows up in 213112 even if the operator’s staff count is steady. For that reason the upstream total is more stable as a measure of employment tied to production than any single line.

Seasonally adjusted series and the series that carry annual averages

BLS publishes each industry twice. Series codes beginning CES are seasonally adjusted; codes beginning CEU are not. For oil and gas extraction those are CES1021100001 and CEU1021100001, and for support activities CES1021311201 and CEU1021311201. The seasonally adjusted versions are the right tool for asking whether employment rose or fell last month. Only the unadjusted versions carry annual averages, which BLS’s calculation chapter describes as the “sum of monthly estimates divided by 12.” Comparing a seasonally adjusted month with an unadjusted annual average mixes two measures, and this page avoids it.

The support activities series carries a flag that the extraction series does not. On data.bls.gov, CES1021311201 is marked “(I)”, seasonally adjusted independently. BLS explains that series published after the first preliminary estimate “are seasonally adjusted directly but are not used in aggregation.” Separately, the detailed support and drilling series appear one month later than the extraction series, which is why Table 1’s 2026 row runs through August for those lines and through September for extraction.

The monthly record: highs, lows and one-month steps

Annual averages hide the turning points. Table 3 takes the seasonally adjusted monthly series from January 2016 to the latest month published and pulls out the readings that define the decade. The extraction series began 2016 at its highest value of the period, because the industry was still shedding staff hired before the oil price collapse that started in 2014. Our inflation-adjusted oil price history shows that collapse in constant dollars.

The single largest monthly move in either series was April 2020, when support activities fell by 33,400 jobs as the collapse in fuel demand that spring halted drilling and completion work; May took off another 17,700. Extraction fell far more gently that spring, by 2,100 in April. Its two biggest one-month moves came later and look different. Between December 2020 and January 2021 the seasonally adjusted extraction series dropped 12,900, from 126.6 thousand to 113.7 thousand. In 2022 it rose 6,400 in April and fell 7,400 in August. Crude production shows no matching jumps in those months, and the BLS series page gives no footnote for them.

Readers comparing single months across those dates should treat the steps with caution. Annual benchmarking is one known source of level shifts in CES data: BLS realigns the sample estimates each year to the March level from the Quarterly Census of Employment and Wages, which supplies about 97% of the benchmark employment, and “all data revised as a result of benchmarking are published in February of each year.” Annual averages and year-over-year comparisons are less sensitive to a step in any one month, which is why the arithmetic in this article uses them.

More oil from fewer workers

Set the employment table beside EIA’s production series and the central fact of the shale decade appears. In 2016 the US produced 8.86 million barrels of crude a day; in 2025, 13.66 million, up 54%. Over the same years extraction-company jobs fell 30.7%. Table 4 divides one by the other.

The arithmetic for the end years, so it can be checked by hand: 13,662 thousand barrels a day × 1,000 × 365 days = 4,986.6 million barrels in 2025. Divided by 117,600 extraction jobs, that is about 42,400 barrels per job. For 2016, 8,859 × 1,000 × 366 = 3,242.4 million barrels, divided by 169,800 jobs, is about 19,100. Output per extraction job rose by a factor of 2.2. On the broader upstream count, which includes the drilling and service crews, the 2025 figure is 4,986.6 million barrels ÷ 372,500 jobs, about 13,400 barrels, or 36.7 barrels a day, against 20.9 barrels a day in 2016.

Barrels per upstream job rose in every year but one. The exception is 2022, when upstream employment grew 11.5% as drilling resumed while crude output grew 6.2%; barrels per upstream job slipped from about 12,600 to 12,000. 2020 and 2021 show the opposite distortion: output held up better than jobs because wells drilled before the pandemic kept flowing while crews were laid off. A falling rig count cuts service jobs within weeks, but production from existing wells declines over months, so the ratio jumps in a bust and sags in a recovery.

The physical reason behind the trend is that each new well does far more than an older one. Longer laterals, more frac stages, and many wells drilled from one pad mean a single rig and crew move fewer times per barrel. Our count of US oil wells by production rate shows how a small share of high-rate horizontal wells now carries most output, while hundreds of thousands of low-rate wells each need a pumper’s visit. And the drilled-but-uncompleted well inventory explains how operators drew down a backlog of wells in 2020 and 2021, producing new barrels with far fewer drilling crews.

Three cycles in ten years

2016 to 2019: service hiring, flat operator staff

The recovery after the 2015 to 2016 bust was carried entirely by contractors. Support activities rose from 205,800 in 2016 to 270,000 in 2019 and drilling from 47,900 to 58,000, while extraction-company staff fell from 169,800 to 143,600 and then held near that level. Crude production climbed 39% between 2016 and 2019. Operators kept their own payrolls lean and bought drilling and completion work by the job, a pattern visible in Table 1 as a widening gap between the support column and the extraction column.

2020 and 2021: the pandemic cut

Upstream employment fell from 471,600 in 2019 to 365,100 in 2020 and 327,600 in 2021, a two-year loss of 144,000 jobs, or 30.5%. Support activities took most of the 2020 loss (down 74,700, against 13,800 at extraction companies). In 2021 the two lines fell by similar amounts: support activities by 18,200 and extraction by 17,600, to 112,200, the lowest extraction average in the table. Refineries lost 7,500 jobs between 2019 and 2021, the steepest fall in that line. Crude output in 2021 averaged 11.32 million barrels a day, 8.2% below 2019.

2022 to 2026: a partial rebound, then a slow slide

Upstream employment climbed back to 390,800 in 2023, still 80,800 short of 2019, and then fell to 380,600 in 2024 and 372,500 in 2025. Crude production set a new annual record in each of 2023, 2024 and 2025. Pipelines moved the other way from the rest of the industry, adding jobs in each year from 2022 to 2025. The extraction line slid from 121,300 in 2024 to 117,600 in 2025 and support activities from 216,700 to 212,200, while drilling held at about 42,700. For the cost side of that decision, see our shale breakeven price analysis, which tracks the oil price operators say they need.

2026 so far against the same months of 2025

Because oilfield payrolls have a seasonal pattern, the fair way to judge 2026 is against the same months of 2025. Table 5 does that with the unadjusted monthly values.

Extraction and support jobs are both lower than a year earlier, by 2.4% and 3.9%, while drilling contractors have added 6.1% and crude output is up 1.9%. Drilling employment, 42,700 in January 2025, reached 45,700 in June 2026, its highest monthly value since December 2023. Taken together, the three upstream lines averaged about 366,600 jobs in January to August 2026 against 375,200 in the same months of 2025, so barrels a day per upstream job rose again, from about 36.1 to 37.6.

Two cautions apply to the latest months. BLS revises recent monthly estimates as more payroll reports arrive, so the preliminary August and September values can move, and the February 2027 benchmark revision will restate the whole of 2026. For Permian-specific activity behind these national totals, see the Permian Basin production outlook for 2026.

Why trade groups report two million, or 19 million

Industry employment estimates differ mainly in two choices: which industries count as “oil and gas,” and whether to add jobs that the industry’s spending creates elsewhere. BLS payroll data answer the first choice narrowly and do not make the second at all. Trade group studies answer both broadly. Table 6 lines them up from narrowest to widest.

The first jump, from 117,600 to 372,500, is the contractor question covered above. The second, to 492,100, adds the midstream and downstream industries closest to oil and gas. The third shows how much a retail definition changes the result: gasoline stations and fuel dealers alone employed more than a million people in 2025, over twice as many as the five production, refining and pipeline industries combined. About 85% of them (889,400) are at gasoline stations with convenience stores, and whether they belong in an “oil and gas jobs” total is a choice of definition, not a fact. TIPRO’s press release does not list its 19 sectors, so this page cannot reproduce its 2,043,859; the BLS lines in the table show that a total of that size requires retail, wholesale or manufacturing sectors well beyond the wellhead.

Direct, indirect and induced jobs

The jump from about 2 million to more than 19 million comes from economic multipliers. A study of this kind starts with direct jobs, then uses an input-output model to estimate indirect jobs at suppliers (steel pipe mills, sand mines, trucking, banks) and induced jobs supported by employees spending their wages (restaurants, retail, health care). TIPRO’s release describes its Texas figure this way: “When incorporating direct, indirect, and induced multipliers for oil and gas employment, the industry supported a total of 2,509,121 jobs in Texas last year,” against 476,777 direct jobs there, a ratio of about 5.3. The national ratio implied by its US figures is at least 9.3.

The American Petroleum Institute has also published “supported jobs” totals from studies it commissioned from PwC, built on the same direct, indirect and induced logic. Those studies use their own industry definitions and data years, and their figures should be compared only with figures built the same way. A supported-jobs total is not a headcount. The jobs it attributes to oil and gas are also attributed, by the same method, to the industries whose spending supports them, so supported-jobs totals for different industries cannot be added together. Nor can a supported-jobs figure be compared with the BLS payroll count; one measures payroll positions in named industries, the other models economic dependence.

State figures and the Texas share

Trade groups also publish state counts, and these are often where conflicting national figures originate. TIPRO reports Texas upstream employment monthly from BLS state data; its October 16, 2026 release put September 2026 at 62,600 jobs in oil and natural gas extraction and 133,800 in the service sector, “subject to revisions,” for a combined 196,400. Set beside the national unadjusted September 2026 extraction estimate of 113,600, Texas accounts for a little over half of US extraction-company jobs, though state and national CES estimates are produced separately and do not necessarily sum. For the production side of the same states, see our US oil production by state.

Where the payroll count misses oilfield work

The BLS upstream total understates the number of people whose work depends on drilling and production. Payroll data leave out several groups, each for a definable reason.

• Self-employed workers. Independent landmen, consulting geologists working alone and owner-operator truck drivers are outside the establishment survey entirely, because it excludes sole proprietors and the unincorporated self-employed.
• Contractors classified elsewhere. Water haulers, sand and proppant trucking, environmental and engineering consultancies, and wellhead equipment makers are counted in their own industries (Table 2).
• Royalty owners and mineral holders. Receiving royalty income is not employment in any survey.
• Petrochemical plants and natural gas utilities. Chemical manufacturing and gas distribution are separate NAICS industries and are left out of Table 1, though some broader definitions include them.

The count can also overstate in one narrow way: multiple jobholders appear once per payroll job, so a worker paid by two service companies in the reference pay period counts twice. In an industry where crews rotate between employers, that effect is real but small relative to the exclusions listed above.

Headquarters staff are a further judgment call. An operator’s accountants and land department in a Houston office are generally classified with the operator, in NAICS 211, when the office serves the company’s own production. A separate holding-company office may be classified in management of companies instead. The industry assignment follows the establishment, and BLS does not publish which large companies fall where.

Using the series yourself

Anyone can rebuild Table 1 from the BLS data tools. The series pages on data.bls.gov take a series code, a year range and an “include annual averages” option; the annual averages appear only on CEU series. The numbers stay consistent if you hold to the following.

1. Use CEU (unadjusted) codes and annual averages for year-by-year tables, and CES (seasonally adjusted) codes for month-to-month moves.
2. Mark the last two months as preliminary, and expect support activities and drilling to lag extraction by one month.
3. Re-download after each February benchmark revision; the whole recent history can shift.
4. When dividing output by jobs, match the year and the coverage: crude and gas for the upstream industries, refinery runs for refinery jobs.

A worked example shows why the first habit matters. The latest seasonally adjusted extraction reading, 114,100 for September 2026, sits 3.0% below the 2025 annual average of 117,600. That comparison sets one adjusted month against an unadjusted twelve-month mean. The like-for-like comparison uses the unadjusted September values: 115,600 in September 2025 and 113,600 in September 2026 (preliminary), a fall of 1.7%. The year-to-date average in Table 5, down 2.4%, is the steadier of the two, because it does not rest on one preliminary month. Three honest numbers for the same question can differ by more than a point, so a quoted change should always carry its basis.

The same discipline applies to sums. The upstream total of 372,500 for 2025 is the kind of figure that circulates in trade press as a round “373,000 upstream jobs”, often without saying that it adds three separate BLS series. Each series has its own sample and its own revisions, so the sum is this page’s construction rather than a BLS published total, and BLS’s broader aggregate, mining, quarrying and oil and gas extraction (NAICS 21), also includes coal, metal and stone mining. Quote the components with the total, and the reader can check both.

Activity indicators move ahead of payrolls. A falling count of working rigs usually precedes a fall in drilling and support employment by weeks to months. PetroEyes tracks the weekly figure on its rig count page, and the US refinery capacity guide covers the downstream plants behind the refinery column. The broader set of supply guides sits in the US oil production and supply hub.

Data and sources: U.S. Bureau of Labor Statistics, Current Employment Statistics (national), all employees, thousands: oil and gas extraction, seasonally adjusted (CES1021100001), not seasonally adjusted (CEU1021100001), support activities for oil and gas operations, seasonally adjusted (CES1021311201), not seasonally adjusted (CEU1021311201), drilling oil and gas wells (CEU1021311101), petroleum refineries (CEU3232411001), pipeline transportation (CEU4348600001), gasoline stations and fuel dealers (CEU4245700001) and gasoline stations with convenience stores (CEU4245711001), monthly values and annual averages 2016 to 2026, extracted October 31, 2026. BLS Handbook of Methods, CES concepts and calculation chapters, checked October 31, 2026. BLS Industries at a Glance, Oil and Gas Extraction: NAICS 211 and Support Activities for Mining: NAICS 213, checked October 31, 2026. U.S. Energy Information Administration, U.S. Field Production of Crude Oil (annual and monthly) and U.S. Natural Gas Marketed Production (annual), release of October 30, 2026, checked October 31, 2026. TIPRO, TIPRO Releases 2026 State of Energy Report (March 6, 2026) and Texas Upstream Employment Softens in September (October 16, 2026), checked October 31, 2026. Sums, year-to-date averages, output-per-job ratios and percentage changes are this page’s arithmetic.