Buprenorphine reduces cocaine self-administration by rhesus monkeys, opiate- and cocaine-dependent men and polydrug abusers, but the mechanisms underlying these cocaine-opiate interactions are not well understood. In the present study, the effects of daily placebo or buprenorphine (0.1, 0.3 and 1.0 mg/kg) treatment on cocaine self-administration (0.001-0.3 mg/kg/inject) were examined in five cocaine-experienced rhesus monkeys. Saline and each of six cocaine doses were available in an irregular order. Responding for cocaine (or saline) and food was maintained on a second order FR4 (VR 16:5) schedule of reinforcement. During placebo treatment, the daily number of cocaine injections increased as the unit dose was increased and then decreased at higher doses. Cocaine doses that maintained the highest rates of responding during placebo treatment were more resistant to buprenorphine's effects. The typical increase in response rate during the first five cocaine injections of a session also was attenuated by buprenorphine. The ascending limb of the cocaine dose-response curve was shifted downward and approximately one log unit to the right during low-dose buprenorphine treatment (0.1 mg/kg/day). In contrast, individual response rates for food pellets were unaffected. We conclude that buprenorphine selectively decreases self-administration of some unit doses of cocaine at doses that have minimal effects on food-maintained responding.