A caffeine versus placebo discrimination procedure was used to determine the lowest caffeine dose that could produce discrimination and self‐reported mood effects in normal volunteers. During daily sessions under double‐blind conditions, caffeine‐abstinent subjects orally ingested a capsule containing 178 mg caffeine or placebo. Before beginning discrimination training, the compounds were identified to subjects by letter codes. Fifteen, 30, and 45 min after capsule ingestion, subjects guessed the capsule's letter code. Correct guesses at 45 min earned money. After each session, subjects received a supplementary capsule containing caffeine or placebo to ensure that, within each phase of the study, subjects received the same daily dose of caffeine equal to the training dose. Five of the 15 subjects acquired the caffeine versus placebo discrimination within the first 20 sessions (≥75% correct); 6 other subjects acquired the discrimination with additional training. Nine subjects who acquired the discrimination were subsequently trained at progressively lower caffeine doses. In general, the lowest dose to produce discrimination (≥75% correct) was also the lowest dose to produce self‐reported mood effects: 4 subjects showed discrimination and self‐reported mood effects at 100 mg caffeine, 2 at 56 mg, 1 at 32 mg, and 1 at 18 mg. One of these subjects also showed self‐reported mood effects at 10 mg. The present study documents discriminative stimulus and self‐reported mood effects of caffeine at doses below those previously shown to affect any behavior in