The study examined the effects of expectation on the subjective effects and oral self-administration of 15 mg d-amphetamine (AMP) and placebo in 40 volunteers who reported no prior use of stimulants other than caffeine. A balanced placebo design was used to create four groups: told Placebo/got Placebo (P/P), told Placebo/got Stimulant (P/S), told Stimulant/got Placebo (S/P), told Stimulant/got Stimulant (S/S). There were three sessions. On one session (INFO), participants received a capsule containing AMP or placebo and were given information about the contents of the capsule according to the balanced placebo design. On another session (NO INFO), participants received no information about the capsule's contents and were given placebo. On the final session, participants were allowed to choose either the INFO or NO INFO capsule. Participants came to the laboratory to ingest their capsules, and then returned to their normal environments where they completed subjective effects questionnaires every 2 h for 8 h. Expectancies influenced the subjective effects reported during the INFO session, regardless of whether subjects actually received AMP or placebo: subjects who expected a stimulant had higher ratings of "feel drug" and "like drug". The pharmacological effects of AMP were also evident on the INFO sessions: AMP produced its prototypic subjective effects regardless of subjects' expectancies. Significant interactions between drug and expectancy were obtained on self-report measures of anxiety and arousal: anxiety was higher for groups who received substances that did not match their expectations (P/S and S/P) and arousal increased most in volunteers who expected placebo but received stimulant. Choice of drug was determined primarily by pharmacology: participants who received AMP on the INFO session usually chose that capsule, regardless of information about its identity (P/S: 8/10; S/S: 9/10). In contrast, participants who received placebo on the INFO session chose this capsule at chance levels, regardless of information about its identity (S/P: 3/10; P/P: 6/10). Thus, expectancy influenced some of the subjective effects of AMP and placebo, but the pharmacological effects of the AMP were instrumental in determining whether volunteers would self-administer the drug.