The purpose of this study was to examine the extent of covariation of subjective and discriminative drug effects as the dose of the discriminated training drug was progressively lowered. Six adult male volunteers with histories of opioid abuse, who were not currently physically dependent, were trained to discriminate the mu-receptor agonist hydromorphone (20 mg, oral) from placebo in daily sessions. They received financial reinforcement for correct responses. The hydromorphone training dose was then progressively reduced (20, 14, 10, 7, 5, and 3.5 mg) while the discrimination reinforcement contingencies remained in effect. Measures of subjective and physiological effects were concurrently collected during each discrimination session. As the training dose decreased, discriminative performance was generally well maintained, although the percent of drug–appropriate responses to hydromorphone did decline from 98% to 75%. The magnitude of the subjective and physiological effects of hydromorphone also decreased as the training dose decreased. At the lowest training dose, there were no physiological effects and few subjective effects of hydromorphone statistically different from placebo, although discrimination behavior remained statistically significant at all doses. These data indicate covariation of subjective effects and discrimination performance and suggest that discrimination behavior may be more sensitive for differentiating among drug conditions than traditional subjective effects measures.