The U.S. Army's claimed Patriot Gulf War success rate fell from a wartime 96% to a later 52% warhead-destruction claim, only 40% of it 'highest confidence'

patriot-missilegulf-warmitverification-methodmilitary-engineeringstatistics

The public reputation of the Patriot air-defense system during the 1991 Gulf War rested on a headline figure. As George N. Lewis and Theodore A. Postol documented in Science & Global Security (1993): "the US Army maintained that Patriot had successfully intercepted almost all of the Scuds (45 out of 47, or 96 percent)." That number did not survive scrutiny. The Army progressively revised the claim downward — and even the revised figure was heavily hedged: "The Army also claims Patriot destroyed 52 percent of the Scud warheads it engaged, but only 40 percent of the claimed warhead kills are in the 'highest confidence' category."

The revision matters on two axes. First, the raw rate roughly halved, from ~96 percent to 52 percent of warheads destroyed. Second, the confidence structure behind the surviving 52 percent was thin: by the Army's own grading, three in five of the claimed kills sat below the "highest confidence" bar. The load-bearing statistic was therefore not "52 percent of Scud warheads destroyed" but the much smaller intersection of claimed-and-high-confidence kills.

The distinction between a wartime number and a number that can be defended after review is the whole subject here. This is the Army's-own-figures leg of the credibility challenge; the independent evidence leg — Lewis and Postol finding no unambiguous video of any warhead kill — is claim-lewis-postol-1993-no-video-evidence-of-patriot-scud-warhead-kill. Both sit downstream of the concrete battery-level failures the vault documents, especially the accumulating clock-drift bug in claim-patriot-dhahran-1991-clock-drift-from-24-bit-truncation-of-tenths. A separately reported and still-unverified GAO figure — that only about 9 percent of claimed kills met a "strongest evidence" standard — would push the defensible rate lower still; it is held open at question-verify-gao-nsiad-92-340-patriot-9-percent-strongest-evidence rather than asserted here.

Update 2026-08-07

The GAO figure this note held open — "still-unverified" pending a primary read — is no longer unverified in that sense. A 2026-08-07 capture confirmed it via GAO's own report abstract (republished by a Justia mirror, quote-checked) and an independent House Government Operations Committee corroboration: "strong evidence exists for about 9 percent of the engagements that the Army believes resulted in warhead kills." See claim-gao-nsiad-92-340-patriot-9-percent-strong-evidence-standard. This pushes the defensible confirmed-kill rate below even this note's 40%-high-confidence figure — the GAO number is a stricter, differently-defined bar (physical/radar corroboration of a specific engagement, not the Army's own confidence grading), so the two numbers are not directly substitutable, but both point the same direction: down from the wartime claim. The open question this note pointed to, question-verify-gao-nsiad-92-340-patriot-9-percent-strongest-evidence, is now marked answered.

Correction history

• 2026-08-14 — source_quote corrected from an ellipsis-joined splice of two separate sentences (the abstract's restatement of the 52% figure, stitched to the introduction's 40%-highest-confidence clause) to the single unbroken sentence in the introduction (p. 2) that actually carries both figures together: "The Army also claims Patriot destroyed 52 percent of the Scud warheads it engaged, but only 40 percent of the claimed warhead kills are in the 'highest confidence' category." No change to the claim's substance — both figures were already correct under the old splice — but the old form implied one continuous sentence where the primary has two, in two different places. Independently re-verified via direct extract_pdf + quote_check against the primary PDF (sha256:51b2ffee...). The same session also independently confirmed, for the first time as its own quoted figure, the wartime "45 out of 47, or 96 percent" baseline this note's title has carried since creation. See question-verify-lewis-postol-1993-patriot-video-evidence-quotes (now answered) and 10-inbox/raw/2026-08-14-do-lewis-postol-1993-state-verbatim-that-the.md.