Baker Hughes Rig Count: What the Friday Number Actually Counts

Every Friday, one number crosses the wire and gets quoted as drilling health in a single line: the Baker Hughes rig count. The headline is usually just the US total and its weekly change - up six, down two, flat. That shorthand hides what the survey actually measures, and it invites the most common error in energy commentary, which is treating a rig move as a production move. A rig count is a census of drilling machines at work in a given week. It is not a count of wells completed, not a count of wells producing, and not a forecast with barrels attached. This guide walks through the published Baker Hughes definitions line by line - what qualifies as a rotary rig, what qualifies as active, what gets left out - then shows how to read the oil, gas, and miscellaneous splits, the trajectory mix, and a weekly change without fooling yourself. The live numbers live on our rig counts page; this page explains the ruler those numbers were measured with.

1. A weekly census, running since 1944

Baker Hughes describes its North American Rotary Rig Count as a weekly census of drilling rigs actively exploring for or developing oil or natural gas in the United States and Canada. The company has issued the counts since 1944, when the Hughes Tool Company began weekly tallies of US and Canadian drilling. A companion International Rotary Rig Count, started in 1975, covers the major producing regions outside North America on a monthly schedule. Baker Hughes states the cadence plainly on its own site: the North American count is released weekly, the international count monthly. That long, consistent run is the reason economists, journalists, and the EIA itself still use the series as the reference history of drilling activity.

The census is compiled by people, not scraped from a database. Baker Hughes field representatives stay in frequent contact with operating rigs in their districts, and regional offices across North America and the other reporting regions gather the activity reports that feed each release. Two consequences follow. First, the count reflects confirmed field activity rather than permits or plans, which is why it is treated as a coincident read on drilling itself. Second, it measures intent that has already become steel on location - a rig only enters the census once it is on site and drilling, as the next section spells out. Permitted wells, contracted rigs, and announced programs do not appear until they pass that test.

2. The definition, word by word

Three terms in the name do the real work: rotary, rig, and active. Each one draws a boundary, and each boundary explains a headline surprise.

A rotary rig drills by rotating drill pipe from the surface to cut a new wellbore, or to sidetrack an existing well into a new target. Baker Hughes includes only rigs that are significant consumers of oilfield services and supplies. Cable tool rigs are excluded. So are very small truck-mounted rigs and rigs that can operate without a permit. The interesting edge case runs the other way: non-rotary units are counted when they are doing rotary work in substance. Coiled tubing units and workover rigs employed in drilling new wells are included in the count. The unit's badge matters less than the job it is doing that week.

Active has a precise field meaning. To be counted, a rig must be on location and drilling - in the phrase the industry uses, turning to the right. It counts as active from the moment the well is spudded (the bit first starts the hole) until the well reaches target depth, often shortened to TD. Everything before and after that window is outside the census. A rig in transit between locations is not active. A rig rigging up - being assembled and tested on the pad - is not active. And a rig that has finished drilling and moved into workovers, completions, or production testing is not active for this survey, even though it is still on the pad and still busy. Section 3 returns to that exclusion, because it is where the rig count and the completion story part ways.

Put together, the Friday number answers a narrow question well: how many qualifying rigs were in the drilling window, on location, during the survey week? It does not answer how many wells were finished, how much pipe was laid, or how much oil will flow. Keeping the question narrow is what makes the series usable across eight decades. The mistakes start when readers stretch it.

3. What the count leaves out

The exclusions explain most weeks when the rig count and the physical market seem to disagree. Four of them deserve a permanent note.

Workover and completion activity on existing wells. A workover rig pulling tubing, cleaning out a well, or re-stimulating an old producer can add real barrels, and a completion crew fracturing a drilled well is the step that actually turns a hole into production. Neither enters the census unless the unit is drilling a new wellbore. The rig count therefore goes quiet exactly when the completion side gets loud - for example, when operators work through a backlog of drilled wells without spudding many new ones. That backlog has its own name and its own data, covered in section 6.

Small and permit-exempt rigs. Cable tool rigs, very small truck-mounted rigs, and rigs that can operate without a permit are excluded by definition. The count tracks the commercial drilling fleet that consumes services and supplies at scale, not every auger turning in a shallow field. For supply analysis that is the right scope; for a county-level jobs story it would not be.

Shut-in and producing wells. The census counts drilling, not production status. A field can restart shut-in wells, reactivate stripper wells, or run enhanced recovery projects with no change in the rig count at all. Supply can move while the Friday number sits still, and section 6 shows why that gap is structural rather than a data failure.

The service intensity behind each rig. The count treats each qualifying rig as one unit. It does not weight by horsepower, crew size, pad configuration, or lateral length drilled that week. A modern pad rig walking between wellheads on a multi-well pad and a single-well rig in a conventional field both print as one. That uniformity keeps the census clean, and it also means the count understates how much drilling capacity each unit now represents - the productivity problem in section 7.

4. How to read the splits: oil, gas, miscellaneous, land, offshore

The headline total is the least useful line in the release. The tables underneath it carry the signal, and Baker Hughes publishes them at a fine grain: by state, province, and basin for North America, by major region for the international count, with splits by land versus offshore, by target (oil, gas, miscellaneous), and by trajectory. Our rig counts page archives the same structure - headline totals, splits, basin detail, and weekly history - so the anatomy below can be checked against the latest release in one click.

Oil-directed versus gas-directed is the first split to read. Rig crews work where the economics point them, and oil-directed and gas-directed activity respond to different prices, different basins, and different contract seasons. A flat US total can contain a five-rig rise in oil-directed drilling and a five-rig fall in gas-directed drilling, which is two stories, not zero. The price context sits on our WTI price page for the oil side and our natural gas price page for the gas side. As a reading habit: name the split that moved before you quote the total that did not.

Miscellaneous is the residual line for counted rigs directed at neither oil nor natural gas, including geothermal work. It is small enough to ignore in most weeks and useful mainly as a reminder that the census covers drilling activity, not just hydrocarbons. The latest release in our archive as this guide was published, for the week ending October 2, 2026, shows the scale plainly: the United States counted 598 rigs in total - 456 oil-directed, 133 gas-directed, and 9 miscellaneous - while Canada counted 216 in total - 149 oil, 66 gas, and 1 miscellaneous - and the worldwide total stood at 1,948. When the miscellaneous line moves by one or two, nothing has happened. When the oil line moves by double digits for several weeks, something has.

Land versus offshore and basin detail locate the move. Offshore rigs are few, expensive, and committed on long contracts, so the offshore line moves slowly and says more about projects sanctioned years ago than about this quarter's prices. Land rigs, especially in the shale basins the tables break out individually, respond within weeks. In that same October 2, 2026 archive, the Permian alone accounted for 270 counted rigs, 269 of them oil-directed - close to three-fifths of the US oil-directed fleet in one basin. A national move concentrated in one basin is a basin story (midstream room, local differentials, one operator's program), while a move spread across basins is a price story. Our deep dive on Permian rig counts and the WTI Midland spread works that basin-level reading in detail.

5. Friday at 1 p.m. Eastern: the release rhythm

The North America count is released on Fridays at 1 p.m. Eastern, noon Central. In weeks shortened by a holiday, the release moves to the last business day of the week. The international count follows monthly. Markets have traded on this rhythm for decades, and two practical habits come with it.

First, the count describes drilling activity during the survey week, gathered by district representatives from the rigs themselves. It is a field census published quickly, not a modeled estimate published slowly. That makes it timely and narrow at the same time: timely because the lag between activity and publication is days, narrow because the census only sees the drilling window described in section 2. Second, Friday publication means the number lands after the week's main US petroleum data - the EIA Weekly Petroleum Status Report on Wednesday - has already set the inventory context. Reading the two together is the standard professional sequence: Wednesday tells you what is in the tanks, Friday tells you how hard the industry is drilling to change it. Our guide to reading the EIA weekly report covers the Wednesday half of that sequence in the same step-by-step style.

One calendar warning belongs here. Holiday weeks compress both the release day and the underlying activity - crews rotate, moves are deferred - and year-end weeks add budget-cycle noise as operators finish annual programs. A weekly change printed in the last release of December deserves even more skepticism than usual, which is the subject of section 8.

6. Why rigs up is not barrels up: lag, DUCs, and the completion gap

This is the section to internalize before quoting the count anywhere. Between a rig entering the census and its well adding supply, three gates have to open, and each gate has its own timing and its own data.

Gate one: finish drilling. The rig leaves the census at target depth, but the well is then just a hole in the ground, cased or waiting for casing. No oil or gas flows from a drilled hole. Gate two: complete the well. Completion - casing, cementing, perforating, and hydraulic fracturing in shale wells, followed by flowback and hookup - is done by different crews than the drilling rig, on a schedule the operator controls. Gate three: first production and ramp. Only after completion does the well start producing, and shale wells then decline steeply from their early rates, which is why new supply must first outrun decline from every older well before total production rises at all.

Wells paused between gate one and gate two have a name: drilled but uncompleted wells, or DUCs. The Energy Information Administration estimates DUC inventories for the major tight-oil and shale-gas regions in its monthly Drilling Productivity Report (DPR), alongside production by region. DUCs matter to rig-count reading because they decouple the two series in plain sight. Operators can raise production for a stretch by completing DUCs faster than they drill new wells - supply grows while the rig count falls. They can also drill ahead of completions and build the DUC inventory - the rig count rises while supply waits. The DPR exists precisely because the rig count alone cannot tell you which of those is happening. Neither pattern is a contradiction; both are the completion gap doing its normal work.

The lag compounds the point. The DPR's headline productivity metric, new-well production per rig, is built by dividing estimated production from new wells by the region's rig count lagged by two months, and the EIA cautions that the metric can turn unstable when rig counts or completions swing quickly. Even the agency that publishes the bridge between rigs and production builds a two-month lag into the bridge and warns readers not to lean on it during fast moves. A Friday rig change is therefore best filed as a statement about drilling activity and operator intent this week, with any production consequence arriving months later, filtered through completion schedules and DUC inventories that the rig release does not report.

7. One rig is not one rig: trajectory mix and productivity

The census counts rigs, not capability, and capability per rig has not stood still. The trajectory split - vertical, directional, horizontal - is where the release quietly records that change.

A vertical well goes essentially straight down into a conventional reservoir. A directional well is steered at an angle to reach a target that is not directly below the rig. A horizontal well turns during drilling to run sideways through the producing rock, exposing a long stretch of formation to the wellbore - the standard architecture in the shale basins that dominate the basin tables. Horizontal and directional drilling together are what the industry means by unconventional development, and in the US count they are where the action is: a basin line like the Permian's 270 rigs in the October 2, 2026 archive is overwhelmingly a horizontal fleet drilling long laterals from multi-well pads.

Two reading rules follow, and both can be stated without inventing a single productivity figure. First, cross-era comparisons of the raw count mislead. A US total that looks low against a decades-old peak still describes a fleet drilling longer laterals, more wells per pad, and more producing rock per rig-week than the fleet behind that peak. The EIA's DPR tracks the consequence directly - new-well oil and gas production per rig, by region - and the long-run direction of that series, not the raw rig total, is the right way to argue that each rig now delivers more. Second, within a single era, the trajectory mix refines the headline. A weekly rise concentrated in horizontal oil-directed rigs in one shale basin promises more future supply per added rig than the same rise spread across vertical rigs in mature conventional fields. The tables give you that mix every week; the headline number throws it away.

Pad drilling sharpens the point. A rig on a multi-well pad drills one well, moves a few feet on the same pad, and drills the next, staying inside the census almost continuously while delivering several wells in sequence. Batch completions then fracture those wells together. One counted rig, several wells, one completion campaign - the arithmetic between rigs, wells, and barrels has three separate conversion rates, and the Friday release reports only the first input. Treat any analysis that multiplies a rig change by a fixed barrels-per-rig figure, without naming its trajectory, basin, and completion assumptions, as arithmetic theater.

8. Noise versus trend: how to read a weekly change

Rigs move. A rig finishes a well on Tuesday, spends days moving and rigging up, and spuds the next well the following week; across a national fleet, those gaps, plus weather, holidays, and reporting timing, make the weekly change jump around even when the underlying program has not changed at all. Operators also add and release rigs in batches at contract boundaries, which turns smooth intent into lumpy prints. The single-week change is therefore the noisiest number in the release, and the most quoted. Reverse that priority.

Work in fours. Add the last four weekly changes together and read the sum as the monthly direction; drilling programs are budgeted monthly and quarterly, so a four-week window matches the decision rhythm that produces the data. Then place the level, not just the change, against its year-ago value, which the release tables and our archive both carry. In the October 2, 2026 archive, for example, the US total fell by 1 on the week while standing 49 rigs above the same week a year earlier, and Canada rose by 8 on the week and 26 year over year. The weekly column says drift; the annual column says expansion. Both are true, and confusing them is how a quiet Friday becomes a false headline.

Three quick filters separate signal from weather. Persistence: the same direction for three or four consecutive weeks, in the same split, is a program change; one week is a schedule. Breadth: a move shared by several basins and both land regions is a price response; a move in one basin is a local story - check that basin's differentials and midstream news before generalizing. Price confirmation: rigs follow prices with a lag, so compare the four-week rig direction with where WTI and Henry Hub have actually traded over the past quarter, not with this morning's tick. If rigs and the trailing price trend disagree, the rigs are usually late, not wrong - contracts and pad schedules commit operators weeks ahead.

9. Worked example: reading a hypothetical "US rigs +6" headline

The example below is a labeled hypothetical - invented teaching numbers built to show the reading order, not a real release. Suppose a Friday headline reads: US rig count up 6 on the week. Here is the walk-through, in the order a careful reader runs it.

Step 1: split the six. In this hypothetical, the tables show oil-directed rigs up 8, gas-directed down 2, miscellaneous flat, with the Permian contributing 5 of the 8 oil rigs and the Haynesville accounting for the gas decline. The story is now an oil story in one basin with gas softening elsewhere - three facts the headline discarded. Step 2: run the four-week sum. Suppose the prior three weekly changes for the US total were minus 2, plus 1, and minus 3. The four-week sum is plus 2. The month reads as flat with a noisy final week, not as a drilling surge. Step 3: check the trajectory and geography. If the added oil rigs are horizontal Permian rigs on multi-well pads, each one carries more future supply than an average rig; if they are vertical rigs scattered across mature fields, each carries less. The hypothetical specifies the former, so the activity signal is stronger than the raw six suggests - and still says nothing about this quarter's production.

Step 4: cross-check the completion side. Nothing in the rig release tells you whether completion crews kept pace. The check lives in the EIA's monthly Drilling Productivity Report - DUC inventories by region and new-well production per rig - and, weekly, in whether the Wednesday EIA report's production estimate is trending with or against the drilling story. In this hypothetical, assume the latest DPR showed Permian DUCs flat: drilled wells are being completed roughly as fast as they are drilled, so the added rigs point to future supply arriving on the normal lag rather than into a growing backlog. Step 5: state what the number cannot say. The correct summary of the hypothetical release is one sentence: US drilling activity ticked up on the week, led by Permian horizontal oil rigs, with the four-week trend flat and any production effect months out and dependent on completions. Anyone who summarized it as "US oil supply rises" has invented two steps of the chain in section 6.

Run the same five steps on a down week and the logic holds in mirror image: split first, smooth second, locate third, complete the cross-check, and refuse the production leap. The checklist in the next section compresses the routine into a table you can keep open next to the release.

10. The Friday reading checklist

Work down the rows in order. Each row names where to look and what the check rules in or out, so a single headline cannot skip the queue.

Rows 1 through 5 come from the release itself and take minutes. Rows 6 through 8 are the discipline: they force the rig number to sit next to completions, prices, and inventories before it becomes a claim about supply. A weekly take that skips rows 6 through 8 may still be a fine drilling story. It is not yet an oil-market story.

11. Five common misreads

Misread 1: "Rigs rose, so production rises next month." The chain in section 6 - drill, complete, produce, outrun decline - takes months, and DUC inventories can absorb or accelerate it. The EIA builds a two-month lag into its own rig-to-production bridge and warns the bridge wobbles when activity swings. File rig news under drilling activity; file production claims under EIA production data.

Misread 2: "The rig count is back to its old peak, so drilling is back." Raw counts across decades compare unlike fleets. Longer laterals, pad drilling, and higher new-well production per rig - the productivity series the EIA publishes by region - mean each modern rig represents more drilling capacity than a rig from an earlier peak. Compare eras with the productivity data or not at all.

Misread 3: "The total was flat, so nothing happened." Check the splits before concluding calm. Oil up and gas down in equal size prints as flat and describes two active decisions. The same applies across basins: a Permian gain offset by declines elsewhere is a reallocation story wearing a flat headline.

Misread 4: "Canada and the US move together." They share a release, not a season. Canadian activity swings hard with winter access and spring breakup, when soft ground restricts moves, on top of the same price signals the US fleet answers. In the October 2, 2026 archive Canada rose 8 on the week while the US fell 1 - a normal divergence, and a reminder to read the two censuses as two censuses.

Misread 5: "One rig equals a fixed number of barrels, so multiply." There is no stable barrels-per-rig constant: it varies by basin, trajectory, lateral length, completion design, and where the well sits in the DUC queue. The EIA publishes new-well production per rig as a modeled, lagged, region-specific estimate and flags its instability during rapid changes. Borrowing a single multiplier from a headline and applying it to a weekly change manufactures precision the source data does not contain.

12. Frequently asked questions

What does the Baker Hughes rig count actually count?

It counts rotary drilling rigs that are on location and actively drilling a new well (or a sidetrack) for oil or natural gas, from the moment the well is spudded until it reaches target depth. Rigs moving between sites, rigging up, or doing workovers, completions, or production testing are not counted. The North America count covers the United States and Canada each week; a separate international count is published monthly.

When is the Baker Hughes rig count released?

The North America count is released on Fridays at 1 p.m. Eastern (noon Central), shifting to the last business day of the week when a holiday shortens the week. The international count follows a monthly schedule. Baker Hughes has published the counts since 1944, which is why the series anchors long-run drilling history.

Does a higher rig count mean higher oil production?

Not soon, and not one-for-one. A rig counted this week is at the start of a chain: finish drilling, complete the well, then first production. Drilled but uncompleted wells can sit in inventory between steps, and each modern rig also produces more new-well output than rigs in earlier decades, which the EIA tracks as new-well production per rig. Read the rig count as drilling activity and intent; read production in the EIA weekly and monthly data.

What is the miscellaneous category in the rig count?

It is the small residual bucket for counted rigs that are directed at neither oil nor natural gas, such as geothermal drilling. In the release archived on our rig counts page for the week ending October 2, 2026, miscellaneous rigs numbered 9 in the United States and 1 in Canada, against 598 and 216 total rigs. A move in the miscellaneous line almost never drives the story; check the oil and gas splits first.

What do vertical, directional, and horizontal mean in the count?

They describe the well path the rig is drilling. Vertical wells go essentially straight down. Directional wells are steered at an angle but are not run sideways through the target rock for a long distance. Horizontal wells turn to run laterally through the producing formation, which is the standard design in US shale plays. The mix matters because one horizontal rig typically drills far more producing rock per well than one vertical rig, so the same headline count describes different amounts of future supply in different eras.

Where can I see the current rig count and its history?

Baker Hughes publishes the release each week, and our rig counts page archives the headline US, Canada, and worldwide totals with the oil, gas, and miscellaneous splits, basin detail, and weekly history. Pair it with the WTI price page for the price signal operators were responding to, and the natural gas price page for the gas-directed side.

13. Related reading

• Baker Hughes Rig Count Today - the live tracker behind this guide: US, Canada, and worldwide totals, oil/gas/miscellaneous splits, basin detail, and weekly history.
• How to Read the EIA Weekly Petroleum Status Report - the Wednesday release that sets the inventory context before Friday drilling lands.
• Permian Rig Count vs. WTI Spread Analysis - the basin-level version of this reading: rigs, pipeline room, and the Midland differential together.
• WTI Crude Oil Price Today - the benchmark US operators respond to when they add or release rigs.
• Natural Gas (Henry Hub) Price Today - the gas-directed half of the count, with the storage picture behind it.
• Crude Oil Prices: WTI, Brent, Henry Hub - front-month benchmarks for the price-confirmation step in the checklist.