Opioid-dependent volunteers were trained in a three-choice drug discrimination procedure to discriminate between the effects of i.m. administered saline (0.5 ml), hydromorphone hydrochloride (10 mg) and naloxone hydrochloride (0.15 mg). Subjects earned monetary reinforcement by correctly identifying the training drugs by letter code. Subjects received a single drug administration in each daily experimental session. During each session, subjective and physiological effects and three behavioral measures of discrimination (including an operant response, a qualitative response and a quantitative response) were collected. The study was done in three phases: a training phase in which training drugs were identified to subjects by letter code before the session, a test of acquisition phase in which the subject's ability to identify the training drug by letter code was tested and a generalization phase in which dose-response curves for the two active training drugs were tested. Results of acquisition testing of the training drug doses were similar in all three discrimination measures. The two active drugs produced contrasting effects on the various subjective report measures (opioid agonist-like effects vs. opioid withdrawal-like effects). In generalization testing, hydromorphone produced dose-related hydromorphone-appropriate responding, and naloxone produced dose-related naloxone-appropriate responding. Lowest doses of each drug produced saline-appropriate responding; there was no cross-generalization between either active drug. Similarly, hydromorphone and naloxone produced orderly dose-dependent effects on subjective effect scales similar to those produced by the training doses.