The simultaneous i.v. administration of heroin and cocaine, called a 'speedball,' is often reported clinically, and identification of effective pharmacotherapies is a continuing challenge. We hypothesized that treatment with combinations of a monoamine releaser d-amphetamine, and a mu partial agonist, buprenorphine, might reduce speedball self-administration by rhesus monkeys. Speedballs (0.01 mg/kg/inj cocaine+0.0032 mg/kg/inj heroin) and food (1 g banana-flavored pellets) were available during four daily sessions on a second-order schedule of reinforcement (fixed ratio (FR)2 (variable ratio (VR)16:S)). Monkeys were treated for 10 days with saline or ascending doses of d-amphetamine (0.0032-0.032 mg/kg/h)+buprenorphine (0.075 or 0.237 mg/kg/day) in combination. d-Amphetamine+both doses of buprenorphine produced an amphetamine dose-dependent decrease in speedball self-administration in comparison to the saline treatment baseline (P<0.01-0.001), but food-maintained responding did not change significantly. d-Amphetamine alone (0.032 mg/kg/h) significantly decreased both food (P<0.01) and speedball-maintained responding (P<0.05). During saline control treatment, speedball unit doses of 0.0032 mg/kg/inj cocaine+0.001 mg/kg/inj heroin were at the peak of the speedball dose-effect curve. Daily treatment with 0.01 mg/kg/h d-amphetamine+0.237 mg/kg/day buprenorphine produced a significant downward and rightward shift in the speedball dose-effect curve (P<0.01) and no significant effect on food-maintained responding. A significant decrease in speedball self-administration was sustained over 10 days of treatment. These findings are consistent with our previous reports and suggest that medication mixtures designed to target both the stimulant and the opioid component of the speedball may be an effective approach to polydrug abuse treatment.