In the first experiment, responding was maintained for food under a fixed ratio (FR) 10 with a 6-min timeout reinforcement schedule. Eticlopride (0.1-1.0 mg/kg i.p.) dose-dependently decreased the number of food pellets obtained, whereas naltrexone (3-30 mg/kg i.p.) did not significantly alter responding. In the second experiment, intravenous self-administration of cocaine (vehicle, 125, 250 and 500 micrograms/infusion), heroin (vehicle, 5.4, 9 and 18 micrograms/infusion) and cocaine/heroin combinations were maintained under a FR10 reinforcement schedule. Cocaine/heroin combinations included the aforementioned cocaine doses combined with 5.4 micrograms/infusion heroin (CH5.4) or 18 micrograms/infusion heroin (CH18). Cocaine/heroin combinations dose-dependently decreased the number of infusions compared with cocaine alone. Eticlopride (0.03-0.3 mg/kg i.p.) decreased self-administration of 125 micrograms/infusion cocaine and increased self-administration of 500 micrograms/infusion cocaine. Self-administration of 250 micrograms/infusion cocaine was increased after 0.03 and 0.1 mg/kg and decreased after 0.3 mg/kg eticlopride. Eticlopride decreased heroin self-administration, an effect which may be attributable to its rate-decreasing effects. Eticlopride partially reversed the downward shift of the CH5.4 group but did not reverse the effect in the CH18 group. Naltrexone (1.0-10.0 mg/kg i.p.) decreased self-administration of 5.4 micrograms/infusion heroin and increased self-administration of 18 micrograms/infusion heroin. Self-administration of 9 micrograms/infusion heroin was increased by 1.0 mg/kg, not affected by 3.0 mg/kg and decreased by 10.0 mg/kg naltrexone. For the CH5.4 and CH18 groups, naltrexone dose-dependently shifted the dose-effect curves toward the cocaine dose-effect curve. Therefore, self-administration of cocaine/heroine combinations can be maintained in the rat and downward shifts in the cocaine dose-effect curve after combination with heroin are mediated through a naltrexone-sensitive mechanism.