Knuth wrote the line, in these exact words, in “Structured Programming with go to Statements,” published in ACM Computing Surveys in December 1974 — it sits on page 268 of the scanned paper. What almost never travels with it is the sentence that follows, in which Knuth insists the critical 3% of code should be optimized.
“Premature optimization is the root of all evil” Donald Knuth, "Structured Programming with go to Statements," ACM Computing Surveys 6, no. 4 (December 1974): 261–301, at 268. Verified by Quotle.info.
Image direction — grounded in this quote’s real context; paste into your image tool
In context
Evoking the world of Donald Knuth — American computer scientist · Stanford professor emeritus. Theme: "Premature optimization is the root of all evil". Atmospheric and restrained, cinematic directional light, a muted period-appropriate palette, fine grain and texture. Compose with generous empty space for the quote and a small credit line. No lettering in the image itself.
Minimalist
Minimalist editorial slide for the quote "Premature optimization is the root of all evil". Clean background, a single muted accent color, generous negative space, subtle paper or linen texture, no literal illustration — let the words carry it. Reserve clear space for a serif quote and a small credit line. No lettering in the image itself.
“Computer Programming as an Art,” Communications of the ACM 17, no. 12 (December 1974) — a hedged variant
We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil. Yet we should not pass up our opportunities in that critical 3%.
Excerpt
Quoted from page 268 of the scanned paper. The preceding sentences describe programmers wasting time on the speed of noncritical code; the following ones urge close attention to critical code — but only after it has been identified.
The exact sentence appears on page 268 of the December 1974 Computing Surveys paper — confirmed in the full scan ↗, under the running head “268 • Donald E. Knuth.”
Knuth used a softer version the same year in his Turing Award lecture, adding the qualifier “(or at least most of it) in programming” — visible in the archived lecture text ↗.
Wikiquote’s sourced entry ↗ carries both 1974 wordings and notes that Knuth called the saying “Hoare’s Dictum” in his 1989 paper “The Errors of TeX,” adding that the attribution to Hoare is doubtful.
Hoare, asked about it in 2004, said he had no recollection of how the quotation came about — reproduced in this provenance write-up ↗.
This is a 1974 journal article, so it is well inside the term of US copyright — nothing here is public domain. The short excerpt above is reproduced for identification and verification, with full citation; longer reuse should go through the ACM or the author.
Rights status is our research, not legal advice — and copyright varies by country and edition. For commercial or published use, confirm with the rightsholder. See terms.
Know a source that supports or challenges this? Add a link →
Fact-check
The credit that keeps going to the wrong man
This one is unusual: the quote is genuine, the author is famous, and the misattribution was started by the author himself.
✕
The most commonC. A. R. Hoare Doubtful
In his 1989 paper “The Errors of TeX” Knuth referred to the saying as Hoare’s dictum — apparently having forgotten he wrote it. Wikiquote records that reference and states plainly that the attribution to Hoare is doubtful ↗.
✕
OccasionalEdsger W. Dijkstra Unsupported
Asked in 2004 where the line came from, Hoare replied that he had no recollection of it and guessed he might himself have credited Dijkstra — a guess, not a source, as the provenance write-up documents ↗.
✕
Right author, wrong quoteDonald Knuth (truncated) Mis-worded
The sentence is almost always quoted alone. In the 1974 paper ↗ it is preceded by “say about 97% of the time” and followed immediately by a warning not to pass up the critical 3%.
Knuth was not telling programmers to stop caring about speed. The paragraph the line comes from is an argument for efficiency — he opens it by complaining that people were then “condemning program efficiency,” and he ends it by insisting the genuinely hot 3% of a program deserves careful attention, once you have measured which 3% it is. The famous sentence is the middle of a sandwich, served without its bread.
The same 1974 essay is where Knuth reports his own measurements of where programs actually spend their time, which is why the “97% / 3%” framing is not rhetorical decoration. Strip it away and the aphorism flips from “measure first” into “never optimize” — a reading Knuth spends the surrounding pages arguing against.
The quote is real. The version most people carry around is missing the half that makes it advice.
Explore
Filed under
More provenance-checked quotes on the same theme, each with its reuse status.
Born 1938 · American computer scientist · Stanford professor emeritus
Donald Ervin Knuth (born 10 January 1938) is an American computer scientist and professor emeritus at Stanford University, best known for the multi-volume The Art of Computer Programming and for creating the TeX typesetting system. He received the ACM Turing Award in 1974 — the same year he published both texts in which this sentence appears. Knuth is a meticulous citer of sources, which makes the central irony here delicious: the one attribution he got wrong was his own.
For writers & researchers
Cite this
Cite this source
Knuth, Donald E. “Structured Programming with go to Statements.” ACM Computing Surveys 6, no. 4 (December 1974): 261–301, at 268.
This verification page
Quotle.info. “Who really said ‘Premature optimization is the root of all evil’?” Last verified 22 Jul 2026. https://quotle.info/who-said/premature-optimization-is-the-root-of-all-evil/