How Much of Russia’s Refining Is Actually Down?

September 3, 2026
Cars queueing at a petrol station in Kaliningrad during fuel shortages in July 2026
Drivers queue at a filling station in Kaliningrad, 2 July 2026, during the fuel shortages that spread across most Russian regions over the summer. Photo: Aleksandr Podgorchuk / klops.ru

Figures below cover January to late August 2026 and are attributed to the organisation that produced them, with the date each was given. Claims by parties to the fighting are identified as such.

On 4 July, Ukraine’s General Staff said Ukrainian strikes had disabled 42.74 percent of Russia’s oil refining capacity. The same day, the International Energy Agency put the figure at more than 20 percent. Both were describing the same campaign.

That gap has not closed since. In mid-August two economists took the argument public and landed a factor of two and a half apart. Meanwhile a different set of numbers, produced monthly from tanker tracking and port records, has been showing something specific and largely unquoted.

The problem is not that people disagree about how much damage the campaign has done. It is that at least four different quantities are being reported under the same word, and only one of them can be checked from outside Russia.

Four quantities, one word

Read the published assessments closely and they are not competing estimates of the same thing. They are estimates of different things.

QuantityWhat it countsCan it be verified externally
Refining capacity disabledNameplate throughput of units taken offlineNo. Depends on damage assessment inside Russia and on which units are counted as offline
Refineries damagedNumber of facilities hit, regardless of size or severityPartly, by satellite fire detection, but says nothing about volume
Crude export capacity haltedTerminal and pipeline throughput unavailablePartly, from port and pipeline data
Product output and loadingsFuel actually made and actually shippedYes, from tanker tracking and destination port records
The first three are the ones quoted in headlines. The fourth is the one that can be counted from outside.

Capacity is the most-quoted and the least checkable. Russian companies attribute outages to unscheduled maintenance rather than strike damage, which removes the distinction between the two from the public record. Exact output figures for individual products are classified. And nameplate capacity is not what a plant actually produces, because refineries rarely run flat out.

There is a precedent that puts a number on that last point. During the strike wave of August to October 2025, Reuters reported, on the basis of three Russian industry sources, that 20 percent of Russian refining capacity was taken offline at the peak. The resulting fall in actual refinery output was 6 percent.

What that ratio does and does not mean. A three-to-one gap between capacity offline and output lost in 2025 does not mean the same ratio holds in 2026. The 2026 campaign has hit more plants, hit them repeatedly, and hit the units that matter most rather than peripheral infrastructure. The point is narrower: capacity disabled and fuel not produced are different measurements, and the difference between them has previously been large. Anyone quoting a capacity percentage as though it were a production loss is making an assumption they have not stated.

The spread of published estimates

>20% ~25% 25–28% 40% 42.74% 65–70% IEA, 4 Jul Reuters, H1 2026 Vakulenko, 17 Aug Stubb, early Jul Ukr General Staff, 4 Jul Shiryaev, 19 Aug 0 15 30 45 60 75 Share of Russian refining capacity described as disabled, percent.
Lighter segments show the upper end of a stated range; the IEA figure is open-ended above 20 percent. The Ukrainian General Staff is a party to the fighting and is marked accordingly. Reuters gave roughly a quarter of capacity for the first half of 2026; Russian Forbes separately reported that 54 percent of refineries had been damaged, which counts facilities rather than volume and is not plotted here.

The two ends of that range belong to a public argument. On 17 August Sergey Vakulenko of the Carnegie Russia Eurasia Center said refining had fallen 25 to 28 percent. Two days later the economic commentator Vyacheslav Shiryaev called that assessment worthless and put the loss at 65 to 70 percent. They agreed to debate; Meduza published the written exchange on 31 August, and it ended without convergence.

Two figures did emerge from it that are more concrete than any capacity percentage. Russian gasoline output in July 2026 ran between 2.25 and 2.65 million tonnes for the month, against 3.5 million tonnes in March. Diesel ran 5.2 to 5.6 million tonnes against 7.7 million. That is a fall of roughly a quarter to a third in the products themselves, which sits below both loud estimates and above the quietest one.

What can actually be counted

Fuel that leaves Russia by sea is loaded at identifiable terminals onto identifiable ships and unloaded at identifiable ports. That makes it the one part of the picture measurable without access to Russian data, and several organisations measure it monthly.

4.7 Mt Russian oil product loadings in July 2026, according to the Centre for Research on Energy and Clean Air — the lowest monthly figure in its records, and less than half the 9.6 million tonnes loaded in July 2025.

The same monthly analysis records product loadings falling 21 percent month on month in June to a record low, then a further 23 percent in July. Revenue from seaborne product exports measured at destination ports fell 45 percent month on month in July, to 116 million euros a day, with volumes down 36 percent.

Cargo trackers show the same direction from a different angle. Vortexa recorded Russian seaborne diesel and gasoil exports falling to a seasonal minimum of 740,000 barrels a day in May, 26 percent below the year before, during a month with at least 14 recorded refinery strikes. Argus put the capacity affected by the six refineries that went into full outage in May at 1.5 million barrels a day, close to 21 percent of Russian primary distillation capacity.

On the domestic side the picture is thinner but not blank. Reuters put Russian fuel production in June at 25 percent below the same month a year earlier, and running roughly 20 percent under domestic demand. That is a production figure rather than a capacity figure, and it sits close to the gasoline and diesel output numbers that emerged from the August debate.

These are not the same measurement as domestic supply, and they should not be read as one. Export volumes fall both because less fuel is made and because more of what is made is held back for the domestic market. Russia banned jet fuel exports temporarily in June for exactly that reason. But as a floor under the argument, they are firmer than anything on the capacity side: they are counted, dated and published by organisations that show their method.

How far is 2,500 kilometres

The campaign’s most quoted single number has the same problem in miniature. On 6 July Ukrainian drones struck the Gazprom Neft refinery at Omsk, Russia’s largest by throughput at 22 to 23 million tonnes a year. The distance was reported four different ways, and each is defensible.

ReportedSourceMeasured from
more than 2,000 kmNEXTAthe border with Ukraine
around 2,500 kmThe Moscow Times, CNBCthe border, or Ukrainian territory
around 2,700 kmReuters, citing Ukraine’s General StaffUkrainian-held territory
more than 3,000 kmUkraine’s Ministry of Defence, Special Operations Forcesthe route actually flown
All four describe the same strike on the same day. Fire Point, which builds the FP-1 drone used, described the flight as a record for strike drones worldwide.

One feature of that list deserves noting. The distance from Ukrainian-held territory is greater than the distance from the internationally recognised border, because Russia occupies part of Ukraine’s east. The further the front line moves west, the larger Ukraine’s own range records become. A number presented as a measure of capability is partly a measure of territory lost.

The more consequential fact about Omsk is not its distance. Ukraine’s Ministry of Defence said so itself in a post on 6 July: “Omsk was the last untouched of Russia’s 11 largest gasoline producers. Not anymore.” The General Staff’s fuller account, published on Facebook the same day, identified the element struck as the ELOU-AVT-11 primary crude unit, rated at 8.4 million tonnes a year. Fire Point’s chief executive Iryna Terekh said afterwards that one facility in Russia’s top ten by capacity remained unhit, the Angarsk petrochemical plant in Irkutsk region.

Even the superlative is disputed. The Russian outlet Astra noted that on some industry data the Kirishi and Achinsk refineries may process more crude than Omsk does. What is not in dispute is the direction of travel: whatever else the strike demonstrated, it ended the phase in which reach was the constraint.

The clearest local evidence arrived within two days, and not through any official channel. Reuters, citing two industry sources, reported that the refinery had halted crude processing. In Omsk itself the Topline chain stopped selling petrol to private customers at all 53 of its filling stations, with price boards showing zeros instead of figures; the company attributed this to a failure at the oil depot affecting shipment and logistics. Russian firms describe strike damage as unscheduled maintenance at the refinery and as a logistics failure at the pump, which is one reason the public record holds no usable series for it.

Eight months in sequence

WhenWhat happenedAttribution
Jan–FebStrikes on the Ilsky refinery in Krasnodar and the Almetyevsk processing unit in TatarstanTrade and open-source reporting
Late MarBaltic export terminals at Ust-Luga and Primorsk hit; about 40 percent of crude export capacity, some 2 million barrels a day, assessed as halted including the Druzhba pipelineReuters calculations from market data, 25 Mar
MayAt least 14 recorded refinery strikes; six plants attacked in May went to full outage; first wave of domestic fuel shortages beginsVortexa, Argus, Reuters
JunMoscow refinery at Kapotnya struck; industry sources doubt a restart before year end. Product loadings fall 21 percent to a record lowReuters; CREA
2 JulFuel shortages reported in at least 78 of Russia’s 83 regions plus occupied CrimeaNovaya Gazeta Europe
6 JulOmsk refinery struck, the last of the eleven largest gasoline producersUkraine’s General Staff; confirmed by the Omsk regional governor
AugShortages in at least ten regions; Orsk refinery declared idle for at least six months; emergency supply measures orderedOrenburg governor Solntsev; Deputy PM Novak, 17 Aug
Selected events only. Strike counts vary by counter: the Financial Times put refinery strikes at at least 194 in the first half of 2026, while Ukraine’s 1st Separate Unmanned Systems Centre claimed 697 targets of all types in the same period.

The economic backdrop is less disputed than the damage figures. Russian inflation reached 6 percent in late June against a 4 percent target, with the central bank naming gasoline prices as a risk. The 2026 growth forecast was cut to 0.4 percent. Gallup polling in late June found 60 percent of Russians saying economic conditions were worsening, the highest reading in two decades of that survey.

The repair race

With the target list at the top effectively exhausted, the campaign’s remaining variable is whether damage accumulates faster than it is undone. Both sides of that question have serious advocates.

The independent oil and gas analyst Boris Aronstein has argued that the scale, coordination and repetition of the strikes leave Russia unable to repair a plant before the next attack arrives, producing what he called the most severe crisis in the sector in recent years. The observable evidence for that is repetition: the Yaroslavl refinery was struck four separate times in May alone, and again in August.

Against it, Vakulenko makes a specific technical claim: most of the damaged units are Russian-made, so replacement is not gated on sanctioned Western equipment. Repairs, on this reading, are slow but not blocked. Other analysts put refinery repair timelines at months rather than years, and Russia has mitigated by importing finished fuel, lowering fuel quality standards and restricting exports.

Both positions predict the same short-term picture and different medium-term ones. That makes the next few months genuinely informative rather than merely eventful.

What to watch

Both the campaign and the Russian response continued past the end of August, so the positions described above are those of the dates given rather than of the moment of reading. Three things will show which of the competing readings was closer.

Monthly
Product loadings, not capacity percentagesCREA and the cargo trackers publish on a fixed cycle with stated method. If loadings recover while capacity claims stay flat, the capacity claims were wrong.
Sep–Oct
Harvest season demandDiesel demand rises sharply with the harvest. Regional shortages during it would indicate the system has no slack left; their absence would indicate it does.
Ongoing
Repeat strikes versus new targetsWith the largest plants all hit, a shift to repeat strikes on the same facilities would confirm the repair race framing. A shift down the size list would suggest diminishing returns.

Sources

Giovanni de Briganti

Editor of Defense-Aerospace.com, a news aggregator covering all aspects of the aerospace and defense industries, principally focused on primary sources (governments, industry, international organizations, think tanks, etc) and selected international media as well as original opinion and analysis articles.

Don't Miss