This experiment tested whether bromocriptine or desmethylimipramine (DMI), both agents used clinically to treat cocaine abuse, could specifically alter behavior maintained by cocaine injections. Rhesus monkeys were trained to press a lever in daily experimental sessions under a three-component multiple schedule of reinforcement. In the first and third components, food was available under a fixed-ratio (FR) 30 schedule. In the second component cocaine (0.025 or 0.050 mg/kg/injection, IV) was available under a FR 30 schedule. Monkeys received continuous (24 h/day) IV infusions of several doses of bromocriptine or DMI. Bromocriptine (0.8-6.4 mg/kg/day) was infused for at least the same number of sessions as was required for responding to decline to low levels when the monkeys were allowed to self-administer saline. DMI (0.8-12.8 mg/kg/day) was infused for a minimum of 3 weeks. In some instances, low doses of bromocriptine decreased responding maintained by cocaine without reducing food-maintained responding, while higher doses of bromocriptine decreased responding maintained by either food or cocaine. However, bromocriptine doses that reduced cocaine intake also caused overt stimulation of locomotor activity. In contrast, DMI, at doses as much as 10 times higher than those used clinically to treat cocaine abuse did not affect responding maintained by cocaine or food. These results indicate that bromocriptine can selectively reduce behavior maintained by cocaine, although apparently by a mechanism other than blockade of reinforcing effects. On the other hand, DMI did not alter the reinforcing effects of either cocaine or food under these conditions.