Preclinical and clinical studies suggest that GABAB receptor agonists selectively decrease cocaine use. The behavioral mechanism for the interaction between baclofen and cocaine in humans is not known, nor have its effects been characterized in individuals dependent on both cocaine and methadone. The objective of this study is to determine how maintenance on baclofen influences smoked cocaine's reinforcing and subjective effects, mood and cocaine craving prior to and after the initiation of cocaine use in cocaine-dependent volunteers with and without concurrent opioid dependence. Nontreatment-seeking volunteers (10 nonopioid dependent; seven methadone maintained), residing on an in-patient research unit for 21 days, were maintained on each baclofen dose (0, 30, 60 mg po) for 7 days. A smoked cocaine dose-response curve (0, 12, 25, 50 mg) was determined twice: on days 3-4 and days 6-7 of each baclofen maintenance condition. Cocaine sessions began with a sample trial, when participants smoked the cocaine dose available that session, and five choice trials, when participants chose between smoking the available cocaine dose or receiving one 5 dollars merchandise voucher. The results show that in the nonmethadone group, baclofen (60 mg) decreased self-administration of a low cocaine dose (12 mg). In the methadone group, baclofen decreased craving for cocaine. In both groups, baclofen decreased cocaine's effects on heart rate. Baclofen did not alter cocaine's robust subjective effects (eg 'High,' 'Stimulated') for either group. The results from this laboratory study appear consistent with clinical evidence showing that baclofen decreases cocaine use in nonopioid-dependent patients seeking treatment for cocaine dependence. The distinct pattern of effects in methadone-maintained participants suggests baclofen may not be effective in opioid-dependent cocaine users.