The reinforcing and subjective effects of morphine were determined in five human volunteers with histories of i.v. heroin abuse. Subjects responded under a second-order schedule of i.m. injection. Under this schedule, every 100 lever presses produced a brief stimulus light [fixed ratio (FR) 100:s]; the 30th completion of the FR 100 requirement turned on the light for 15 min and the subject received an i.m. injection of morphine [FR 30 (FR 100:s)]. Once each weekday morphine or placebo was available under this schedule. Each drug dose was available for 1 week. Under these conditions placebo did not maintain responding; 3.75 mg of morphine maintained responding in four of five subjects, and higher morphine doses (7.5, 15 and 30 mg) maintained responding in all five subjects. Subjective effects were measured concurrently: these included measures of drug liking, the Morphine Benzedrine Group scale of the Addiction Research Center Inventory, drug detection and identification. Subjects did not report subjective effects different from placebo for the lowest dose of morphine; the intermediate doses of morphine produced inconsistent effects, and the highest dose of morphine occasioned reports of drug liking and "dope" identifications. These results indicate that there can be a significant dissociation of the reinforcing and the subjective effects of opioids, which has implications for theories of opioid abuse, particularly those assuming that the reinforcing effects are causally related to the euphoric effects of opioids. Furthermore, these results confirm that measures of reinforcing effects and measures of subjective effects do not necessarily lead to identical predictions when used to assess the liability for abuse of a substance.