Room to drill

How the count is made

1. The units. Every drilling spacing unit in the DMR footprint layer whose reported formation is Bakken, with its acreage measured on the map (EPSG:5070): 10,572 units.
2. What is already drilled. The surveyed laterals of Bakken-system wells (wells whose Well Index pools name a non-Bakken pool are left out; blank or confidential pools are kept as unknown), by the feet of lateral inside each unit; a well is present in a unit with at least 660 ft inside it. Footage is expressed in two-mile equivalents (10,560 ft = one).
3. The local ceiling. Mature units (three or more wells and no spud for three years) show what a neighbourhood has actually drilled: its achieved density in two-mile equivalents per 1,280 acres. The target for a unit is the larger of the 75th percentile of the mature units within 10 miles and the median of the units with four or more wells that received a spud in the last three years, in the same neighbourhood. At least eight reference units must lie within 10 miles; otherwise 20 miles, otherwise the county, otherwise the basin as a whole. The county and basin stand-ins are flagged on the map and in every table.
4. Room per unit. Target density × eligible acres / 1,280, less the two-mile equivalents already in the unit and less any candidate footage a smaller overlapping unit has already placed on the same land, divided by the unit's own lateral length and rounded to a whole lateral. Units are processed smallest first, so land is counted once.
5. Placement. New laterals run in the direction of the laterals already in the unit (aligned to the unit's sides when within 20 degrees; the long axis of an empty elongated unit; the nearby laterals for an empty square one). They go into the widest gaps between the existing laterals so that the smallest spacing is as large as possible, and never closer to a neighbour than the larger of 330 ft and 0.6 of the unit's natural spacing at the target. Setbacks are 250 ft from the unit ends (existing laterals stop a median 233 ft from them) and 500 ft from the sides, 250 ft where units overlap. A lateral is at least 5,000 ft; pieces longer than four miles are split.
6. Order headroom is read separately from the latest signed NDIC order linked to each unit: wells allowed, less the surveyed Bakken wells assigned exclusively to the unit, less pending permits by surface location. These facts are extracted from the orders' operative text by machine and are not human-reviewed.

Why the 75th percentile

The first version of this method used the neighbourhood median as the ceiling. Run as of October 2023 and checked against the three years of drilling that followed, that ceiling was exceeded by 61 percent of the units that were drilled. The target used here was chosen among eight candidates on that same hindcast, so the figures that follow are in-sample and flatter it somewhat. 18 percent of the drilled units ended above it, and the drilled units ended at a median 0.80 of it. It flagged 8,900 units as having room: 20.1 percent of those received a new lateral in the three years, against 5.9 percent of the units it did not flag, so a flagged unit was 3.4 times as likely to be drilled. As a ranking of which units get drilled it scores an AUC of 0.67, where 0.5 is no better than chance and 1.0 is perfect. Because it flags most units, nearly every new lateral fell in a flagged unit; that share reflects how many units are flagged more than how accurate the ceiling is. The chart in the next section shows the values by county.

Both numbers

This count comes from a November 2026 study of the same question. The study also classed each unit by an economic horizon (which of these laterals would pay at a $55, $65 or $75 oil price) and applied 1,019 expert adjustments to individual units; with that screen and those adjustments it counted 14,324 candidate laterals, and 15,774 with the screen alone. This page removes both and counts 21,805. Most of the difference is the units the screen left out: those with fewer than 15 analog wells within 20 miles, and those it classed above $75. The economic horizon is not published yet. It will be recomputed from the Division's free monthly production reports, the source behind the well index's production panels, and added here when it is.

What this count is not

• Not permits, orders, leases, plans or a forecast. A candidate is a line that fits a geometric rule. Whether anyone can or will drill it depends on rights, orders, benches, interference, prices and operating constraints that are not in the public layers used here.
• No economics yet. The study behind this page screened units by the price at which nearby wells pay back; that screen is held back until it has been recomputed from the free monthly production reports. Until then every eligible unit is projected, including fringe units whose nearby wells may not support the density.
• Unknown targets. Middle Bakken and Three Forks wells plot on top of one another in plan view, and no public layer says which bench a lateral is in. A candidate is never placed over an existing lateral, so stacked development is under-counted where it happens.
• County and basin stand-ins. Where fewer than eight developed units lie within 20 miles, the county or basin density is used. 5,220 candidates in 1,015 units rest on that stand-in; they are flagged everywhere and drawn lighter.
• Overlapping units are not additive. Land is counted once, so the count in a unit depends on the smaller units that overlap it. Per-unit figures on the well index say so.
• Machine-read orders. Allowances, assigned wells and pending permits come from the order pipeline's reading of signed orders and the DMR layers; they are not human-reviewed and a stated allowance is what that order allowed, not proof that it governs today.
• Survey coverage. Only surveyed horizontal wells count as drilled footage. A well without a survey, a vertical well or a well whose survey ends short of its lateral is not seen, so some room is already taken.
• A snapshot. Wells and surveys as of October 17, 2026; orders through October 21, 2026. The count will be refreshed with the site's data releases and re-tested against the drilling that follows.

Data and sources

Everything on this page is in these files, generated by the same run (JSON; the candidate set as GeoJSON lines in WGS84):

• candidates.geojson: every candidate lateral with its unit, county, township, operator, length, class and target density (21,805 features)
• by-county.json, by-township.json, by-operator.json, headroom-by-county.json
• density-ceilings.json, long-laterals.json, summary.json (the headline counts and both numbers)

Sources: North Dakota Industrial Commission, Department of Mineral Resources, Oil and Gas Division: the public GIS downloads of October 17, 2026 (horizontal surveys, wells layer, drilling spacing units), the Well Index of October 28, 2026 (pools), and signed hearing orders through October 21, 2026 as read by this project's order pipeline. County lines: U.S. Census Bureau TIGERweb. The Division notes that its GIS information may be inaccurate or out of date and does not warrant it. No production data, no subscription content and no order document is used or redistributed here. The method is described on the methodology page; the same candidate set is drawn in the 3D viewer and counted on every spacing-unit page of the well index.