Proven Oil Reserves by Country

The ledger

Proven crude oil reserves, year-end 2025, in billion barrels. Source: OPEC Annual Statistical Bulletin 2026, Table 3.1 (published 2026).

Reserve life — the reserves-to-production ratio — is reserves divided by one year of production at the stated rate. It is a units conversion, not a forecast: no country produces at a flat rate until its last barrel, and every one of these numbers would change tomorrow if either input changed. Two clocks are running inside these two columns and they are not the same clock. Reserve life at September 2026 output uses that month's production, which is depressed by a war: the IEA puts more than 10 mb/d of Gulf output shut in that month, so the Gulf states' figures in that column are inflated by the disruption and should not be read as a normal-conditions life. Reserve life at sustainable capacity uses sustainable production capacity — what the IEA judges reachable within ninety days and holdable — and is the more stable of the two. United States production is a calendar-year 2025 average, not a September 2026 reading. Countries the OPEC+ production table does not itemise carry n/d in both columns rather than an estimate.

Canada's 5.080 billion barrels is the figure as the OPEC bulletin publishes it, and the bulletin's own footnote says it excludes oil sands. Almost all of Canada's recoverable oil is oil sands. Other compilations that include them rank Canada third or fourth in the world at roughly 160 billion barrels. This page prints the source's number and the source's footnote rather than silently substituting a different basis, but a reader using this table to rank Canada is using it wrong.

Reserves and the ability to produce them have come apart

The four largest Middle Eastern holders — Saudi Arabia, Iran, Iraq and Kuwait — hold 717.3 billion barrels between them, 45.6% of the world total. In September 2026 they produced 15.83 mb/d against sustainable capacity of 23.66 mb/d. The gap is 7.83 mb/d of capability that exists on paper and did not reach the market.

That number used to reconcile with the figure the IEA states from the top down, and this month it stopped. On the August 2026 table the same four country rows gave a bottom-up gap of 9.91 mb/d against a stated 8.3 mb/d of shut-in Gulf output — two independent routes to one figure, which is the closest thing a reference page gets to a proof. For September 2026 the IEA states more than 10 mb/d and these four countries account for 7.83. The bottom-up route now falls at least 2.17 mb/d short of the top-down floor. The Institute does not close that gap by arithmetic. What it most plausibly says is that the shut-in has spread past the four largest reserve holders — the EIA's own country breakdown for the same month names seven affected producers, including Qatar and Bahrain, neither of which appears on this table. It may also say that a floor and a sum are not comparable objects, which is the same caution the conflict note on the hub page raises about the IEA and EIA figures themselves.

What survives the loss of the reconciliation is the finding it was used to support, and that finding did not depend on it: the world's reserve ranking and the world's production ranking are now measuring different things. Reserves say who owns the oil. They say nothing about who can move it this quarter.

The United States is the clean illustration in the other direction. It sits ninth on this table with 41.909 billion barrels — 2.7% of the world's proven reserves — and produced 13.7 mb/d in 2025, the largest annual crude output any country has ever recorded. Its reserve life is 8.4 years. Venezuela sits first with 303.701 billion barrels, produced 1.16 mb/d, and carries a reserve life of 717 years. Neither number describes a country running out. They describe two entirely different businesses: one that replaces its reserve base continuously through drilling and books only what it can currently produce, and one that holds an enormous inventory it is not extracting.

What a proven reserve is, and why you cannot get a recovery rate out of it

The most common misreading of a table like this one is that it measures how much oil is in the ground. It does not, and the definition says so. The clearest public drafting of it is the U.S. Securities and Exchange Commission's, which defines proved reserves as “those quantities of oil and gas, which, by analysis of geoscience and engineering data, can be estimated with reasonable certainty to be economically producible — from a given date forward, from known reservoirs, and under existing economic conditions, operating methods, and government regulations.” Source: U.S. Securities and Exchange Commission, Modernization of Oil and Gas Reporting, Release Nos. 33-8995 / 34-59192, adopted December 31, 2008, amending Rule 4-10(a) of Regulation S-X; effective for fiscal years ending on or after December 31, 2009.

Three clauses in that definition do the work. “Reasonable certainty” is a confidence threshold, not a volume: where a probabilistic method is used the SEC requires at least a 90% probability that the quantity recovered equals or exceeds the estimate, so a proved figure is deliberately a low estimate. “Known reservoirs” excludes everything undiscovered. And “existing economic conditions” ties the number to price — the SEC requires an unweighted arithmetic average of the first-day-of-the-month price over the preceding twelve months — which is why a reserve figure can rise in a year with no drilling at all.

The table on this page is not itself filed on that basis. It is the OPEC statistical bulletin's compilation of national reporting, and most of the largest holders are state entities that do not file with the SEC and are not independently audited. The SEC rule is quoted here because it is the most precisely drafted public definition of what a proved reserve is, and because the distinctions it draws apply whoever is doing the reporting. Where a national figure runs on a looser standard, the direction of the difference is not knowable from the outside, which is a further reason not to compute anything from it.

So the number in the reserves column is the remaining volume, at one date, that is economically producible under one set of conditions. Everything else is outside it: the oil already produced, the oil in known reservoirs that cannot be produced at the reference price, the oil in reservoirs nobody has found, and the oil that better technology might one day lift but which no engineer can today book with reasonable certainty.

That is what makes the next question unanswerable from this page. A recovery factor is the share of the oil originally in a reservoir that is ultimately produced — cumulative production divided by original oil in place. Neither term is in the table. Cumulative production is not, because reserves are stated from a given date forward. Original oil in place is not, because it is a geological estimate of the whole accumulation and a reserve is an economic estimate of part of the remainder. Dividing one column here by another does not approximate it; the two quantities a recovery factor is made of are simply absent. Nor can the gap be closed from elsewhere: no comparable per-country series of original oil in place exists, and most of the largest holders on this table publish no estimate of it at all. This reference therefore prints no recovery-factor column. It could only be fabricated, and a fabricated column is worse than a missing one, because a reader cannot tell which of the two they are looking at.

What is published, and how far apart the published figures sit

Measured: see each row; the freshest of these is 2018. Each figure retrieved in full from the publisher named beside it. Source for the first five rows: IEA, “Whatever happened to enhanced oil recovery?”, commentary by Christophe McGlade, Glenn Sondak and Mei Han, published November 27, 2018, together with the IEA World Energy Outlook 2013 estimate it cites.

Read down the What it is not column before reading across any row. The six figures are not six attempts at one number and they do not straddle a consensus. They are five distinct quantities across six rows, on bases that do not convert into one another — a stage in the conventional production sequence, its complement (the same measurement stated the other way round, which is why six rows carry five quantities), a single field's outcome, a theoretical global ceiling, a realised share of supply, and an undiscovered-resource assessment that excludes reserves by construction. Four of the six rows are shares and two are volumes, and no arithmetic turns one kind into the other, because the shares do not share a denominator. Anyone quoting two of them against each other as a range is comparing objects that were never on the same scale.

The disagreement that does matter is visible in two adjacent rows. The IEA's World Energy Outlook 2013 put the theoretical enhanced-recovery prize at up to 300 billion barrels — 19.1% of the 1,572.459 billion barrels of proven reserves in the table above, though the two are measured twelve years apart, the enhanced-recovery estimate in 2013 and the reserve table at year-end 2025, and they are not the same kind of object. That ratio is a sense of scale, not a finding. Against it, the realised contribution of enhanced recovery has been roughly 2% of world supply. The distance between a theoretical 300 billion barrels and an actual 2% is the whole subject, and it is a question about cost and engineering rather than about geology.

How to use this page

Click any column heading to sort; click it again to reverse. Sorting by share of world shows concentration. Sorting by reserve life at sustainable capacity separates the countries producing near their limit from those sitting on inventory. Sorting by region shows how much of the total is Middle Eastern, and the answer is more than half once the United Arab Emirates and Qatar are added to the four above.

The United Arab Emirates sits fifth with 120.000 billion barrels and carries n/d in both reserve-life columns. That is not an oversight on this page: the OPEC+ production table published for September 2026 does not itemise a UAE line, and its listed components fall short of its own stated total by roughly the amount a UAE row would occupy. Rather than back out a figure by subtraction and present it as measured, the cell is empty and this paragraph explains why. A reference page that fills its own gaps by arithmetic stops being a reference page.

The rest of the reference