Each of the seven Colorado River basin states can veto and none can decide, and the cuts are made administratively — the Lake Powell release for water year 2026 was reduced from 7.48 to 6.00 million acre-feet

What is happening?

The 1922 Colorado River Compact apportioned 7.5 million acre-feet a year to the Upper Basin and 7.5 million acre-feet to the Lower Basin, with up to 1 million acre-feet more available to the Lower Basin, and a 1944 treaty added 1.5 million acre-feet for Mexico. Those shares rest on an assumed river of about 15 million acre-feet, which a 2026 legal analysis describes as three to four million acre-feet more than the river actually carries. The imbalance is not only historical: the Department of the Interior reports average inflows of 12.9 million acre-feet over 2000 to 2024 against average consumptive use of 13.1 million acre-feet over 2020 to 2024.

Storage shows the result. On 2026-08-02 Lake Powell stood at 3,522.04 feet and 23 percent full, Lake Mead at 1,040.96 feet and 27 percent full. Total Colorado River system contents were 18,904 thousand acre-feet, or 32 percent of capacity, against 22,744 thousand acre-feet and 39 percent one year earlier. The forecast water year 2026 unregulated inflow to Lake Powell is 3,501 thousand acre-feet, 36 percent of normal.

The seven basin states spent more than two years failing to agree on what replaces the rules that lapse this year. Reclamation published a Final Environmental Impact Statement on 2026-07-31 without a consensus agreement in hand, and a Record of Decision is expected within weeks. The federal government is now setting terms the states could not settle among themselves.

Whose problem is this?

The parties that must absorb the reductions are no longer the ones writing the rule. Consensus among the seven was the only route by which those bearing the cost would also be those choosing it, and that route closed.

Where does this problem end?

The boundary matters because the shortfall is not a measurement dispute. Both the allocated quantity and the delivered quantity are published by the same federal agency, and the disagreement is about who absorbs the difference.

What is the state now, and what should it be?

Now

What it should be

Reclamation names an operational floor rather than a balance target. The July 2026 study states that Reclamation will consider all tools available during the interim period to avoid Lake Powell declining below 3,500 feet, and that monthly release adjustments were made to protect a target elevation of 3,525 feet. The preferred alternative in the Final Environmental Impact Statement sets an adaptive framework through 2036, with annual releases from Lake Powell ranging from 5.0 to 12.0 million acre-feet, Lower Basin reductions of up to 3.0 million acre-feet, and Upper Basin voluntary conservation of up to 200 thousand acre-feet.

The two sources give different expiry dates for the current rules. The Reclamation post-2026 page states that the governing agreements are scheduled to expire at the end of 2026. The Colorado Sun reports the expiration as 2026-09-30. Both were read on 2026-08-07 and the difference is not resolved here.

How big is it?

Reclamation states that between 35 and 40 million people rely on Colorado River water for some, if not all, of their municipal needs. That range is the affected population recorded in the accompanying population file. The Department of the Interior separately states that the river provides water for more than 40 million people, and neither source explains how its own count was built, so the two figures cannot be ranked as nested denominators here.

Agriculture takes the larger share of this water, and the farmland served covers 5.5 million acres across the basin. Reclamation states that agricultural uses depend on 70 percent of Colorado River water.

The narrower number is the one that could not be derived. How many people face an actual curtailment of delivery under the shortage tiers, as opposed to relying on a river that is over-allocated in aggregate, is not published in the sources read here.

Under what conditions does it arise?

The river is governed by a rule that requires agreement among seven states, and no single party can impose a settlement on the others.

1. The allocation is a legal entitlement, not a forecast. The Compact divides fixed annual shares. The river delivers what the snowpack gives, and in water year 2026 the inflow forecast is 36 percent of normal. 2. Agreement needs everyone and each party can refuse. The Upper Basin states opposed mandatory reductions while the Lower Basin states demanded that all states take mandatory cuts, and the two positions have held since early 2024. 3. Storage absorbs the gap, so the failure is quiet. Use exceeding inflow does not interrupt anyone. It shows up as elevation falling by fractions of a foot per week, which is a number in a weekly report rather than an event. 4. The federal government has a fallback and the states know it. Because Interior can act through the environmental review process, no state faces the pure choice between agreement and chaos, which weakens the pressure to concede.

What has been tried?

Every attempt above either moves water between reservoirs or reduces a release. None of them changes the ratio of allocated water to delivered water, which is the quantity that the seven states were negotiating over.

What was found?

Why is it still unsolved?

Coordination failure — every party can veto, none can decide, and the cost of continued disagreement lands on a reservoir rather than on any of the negotiators.

Each state can block an agreement it dislikes, so the reachable outcome is delay rather than a settlement that binds everyone. The Upper Basin states opposed mandatory cuts and offered voluntary conservation in their place, and a law review account attributes that stance to reluctance to accept cuts combined with electoral accountability to the populations of those states. The Lower Basin position is that reductions should be specific and measurable in every state rather than borne by the Lower Basin alone. Both positions are internally coherent, and neither side gains by conceding first, because the concession is permanent while the hydrology that justified it may reverse.

The structure rewards waiting in a second way. Interior can act unilaterally through the environmental review process, and it did, which means no party ever faced the choice between agreeing and having nothing at all. A federal fallback that exists to prevent paralysis also removes the deadline pressure that would end it. The Final Environmental Impact Statement of 2026-07-31 explicitly preserves the opportunity for the basin to keep working toward consensus and to fold any later agreement into future operations, which converts the missed deadlines into a postponement rather than a loss.

Meanwhile the shortfall is absorbed by storage. When use exceeds inflow the difference comes out of Lake Powell and Lake Mead, and no water user is turned away in the year the imbalance occurs. The consequence is deferred until an elevation threshold is crossed, and thresholds such as 3,525 feet and 3,500 feet are themselves administrative choices that can be defended by moving water down from Flaming Gorge. Each such move buys time and reduces the urgency that would otherwise force a settlement.

What observation would mean it is solved?

Candidates — (a) combined Lake Powell and Lake Mead storage rising over several consecutive years (b) basin consumptive use at or below inflow on a common multi-year average (c) a signed consensus agreement among the seven basin states incorporated into federal operations.

None of these three candidates is sufficient on its own, and each can improve for reasons that have nothing to do with the underlying imbalance.

(a) is the weakest. Storage rose in past wet cycles without any change in allocation, and a single strong snowpack year would satisfy it. It measures weather more than governance. (b) is the closest to the real quantity, but the two averages currently published use different windows, 2000 to 2024 for inflow and 2020 to 2024 for use, so a common basis would have to be established before the test could be applied honestly. (c) can be satisfied by an agreement that defers the hard number, which is precisely what the adaptive framework through 2036 permits. An accord that sets a process for deciding later would count as consensus while leaving the allocated-to-delivered gap untouched.

What is it connected to?

Fills with researchthe relation to Glen Canyon Dam hydropower capacity, to groundwater substitution in Arizona and California, to tribal water rights settlements, and to agricultural commodity markets in the Imperial and Yuma valleys was not researched. Relation type and evidence grade are unconfirmed.

What these sources do not say

• What the Record of Decision actually orders. As of 2026-08-07 it had not been issued. The Final Environmental Impact Statement identifies a preferred alternative, but the binding document and its effective date were not available to read.
• How the reductions are apportioned among Arizona, California, and Nevada. The figure of up to 3.0 million acre-feet is a Lower Basin total. One press account reports Arizona as hit hardest, but no state-by-state schedule was confirmed in any source read here.
• Whether current combined storage is an all-time record low. A press report was cited to that effect in search results, but the article returned HTTP 451 and could not be read, so the claim is not carried in this dossier.
• What happens if the Upper Basin simply does not deliver. The legal analysis describes a possible compact call and prepared litigation, but no source read here sets out the mechanism, the forum, or the remedy.
• The gap between paper allocation and actual delivery as a single audited number. The Compact shares of 7.5 million acre-feet to each basin, the additional million available to the Lower Basin, and the 1.5 million acre-feet treaty obligation to Mexico are nowhere published as one total set against the 12.9 million acre-feet of recent average inflow. The size of the paper-to-river gap therefore has no audited figure in the sources read here.
• Why the two expiry dates differ. Reclamation says the agreements expire at the end of 2026 and a newspaper says 2026-09-30, and neither explains the other.

See the evidence

All eight sources above were opened and read on 2026-08-07. Two of the Reclamation documents were retrieved as PDF files and read directly rather than through a summary. Two figures were confirmed in two independent places: the Powell and Mead elevations of roughly 3,522 feet and 1,041 feet appear in both the Reclamation weekly report and the Colorado Sun account, and the combined Powell and Mead storage of 12,420 thousand acre-feet computed from the weekly report is consistent with the 12.7 million acre-feet the newspaper reports for mid-July, since storage was falling through that period. Three cautions belong on the record. First, the weekly report URL serves whatever week is current, so a later reader will see different numbers at the same address. Second, the inflow and consumptive use averages published by Interior use different windows, 2000 to 2024 and 2020 to 2024, and the 0.2 million acre-feet difference between them should not be read as a precise annual deficit. Third, the expiry of the current rules is given as the end of 2026 by Reclamation and as 2026-09-30 by the Colorado Sun, and this dossier reports both rather than choosing. Neither the Final Environmental Impact Statement itself nor the underlying 1922 Compact text was read in full. This is a research-based definition, so observation_refs is empty and provenance_mode is press-derived.

People affected

Estimated range 35,000,000–40,000,000 As of 2026

Sensitivity The interval is not a confidence interval. It is the spread between the two numbers a single agency uses to describe one quantity, and no distribution or sampling error is attached to either endpoint. Reclamation gives 35 to 40 million for reliance on some or all municipal needs, while the Department of the Interior press release of 2026-07-31 states that the river provides water for more than 40 million people. Neither source states how its own count was built, so the two figures cannot be ranked as nested denominators and the larger one is not treated here as an upper bound. The interval is anchored to the Reclamation statement because that is the one carrying an explicit scope. The narrower quantity could not be derived at all: how many people face an actual curtailment of delivery under the Lake Mead shortage tiers, as distinct from depending on a river that is over-allocated in aggregate, is not published in any source read here. Reclamation does publish calendar year 2026 diversions of 0.916 million acre-feet for Metropolitan Water District, 0.951 million acre-feet for the Central Arizona Project and 0.189 million acre-feet of Nevada consumptive use above Hoover, but converting acre-feet into people would require a per-capita use assumption that no source supplies, so it was not attempted.

Regional breakdown No source read here gives population served by state alongside a basin total on the same measurement basis. The three diversion figures that are published by state or project are volumes of water rather than counts of people, and they cover only part of the Lower Basin. Splitting a basin-wide total across the seven states by their resident population would be proportional allocation, which is forbidden and is especially baseless here because reliance on the river does not track state population. Large parts of Colorado and Utah lie outside the basin, while Nevada draws almost all of its municipal supply from it.