Ukraine strikes the plant behind up to 40% of Moscow's petrol for a second time in three months

The two published counts do not cover the same window. Russian authorities put 450 interceptions in the vicinity of the capital, which is 28 percent of the 1,600-plus figure ^[14], and that larger number covers multiple defensive perimeters and reaches back to the previous day ^[11]. The Defense Ministry's 1,110 covers all of Russia on September 20 alone ^[7]. The difference between the two is about 490 drones ^[9]. Both are counts of drones the defender says it downed ^[10][7].

For anyone trying to model petrol supply in Moscow, the useful number is throughput. Crypto Briefing describes the Gazprom Neft-owned Moscow Oil Refinery in Kapotnya as struck, with warehouses and other infrastructure damaged, and gives no throughput figure or outage duration ^[6]. The "up to 40%" share of the capital's petrol ^[15] therefore describes how concentrated supply is. It is not a figure for supply lost. Two people were killed and roughly 20 injured in the broader Moscow region ^[1], on the final day of Russia's parliamentary, regional and local elections ^[4].

The same Kapotnya plant was damaged in June 2026, in an attack involving nearly 200 drones ^[12]. The up-to-40 percent share is attached to it again in September ^[15]. Either the plant was repaired in the interval or the share figure predates both hits, and drone numbers rose eightfold over roughly three months ^[13].

Zelensky said the targets were oil and logistical facilities, and confirmed that domestically manufactured Flamingo and Pelican missiles flew alongside the drone fleet ^[5]. Crypto Briefing's reading is that Ukrainian-made munitions give Kyiv more political freedom to strike deep inside Russia without the restrictions attached to NATO-origin weapons ^[16]. On the largest night the capital has seen since the war began ^[11], the stated target set was oil and logistics ^[5].

In my view the escalation this record establishes is in volume and in domestic production of the delivery systems; a plant hit twice in roughly three months and still credited with up to 40 percent of the capital's petrol ^[15][12][13] is thin evidence of lost refining capacity. Three readings stay open: the damage was superficial and repaired within days, leaving the share intact; it was real but other refineries covered the shortfall, so the capital queues without running dry; or the interception tallies, both of them from Russian sources ^[10][7], overstate what was flown, in which case the eightfold rise is measuring the defender's own counting. Downtime would decide it, and a Kapotnya plant running below capacity into October would make the drone count a fair proxy for effect.