The purpose of the present study was to examine the effects of the 5-HT3 antagonists ondansetron and MDL72222, and the 5-HT releaser and reuptake inhibitor dexfenfluramine, on intravenous heroin self-administration by Wistar rats. Using separate squads of animals, two separate schedules of heroin reinforcement were used; a relatively low dose (0.03 mg/kg per infusion) made available under a FR5 schedule for 1 h each day, and a moderate heroin dose (0.1 mg/kg per infusion) available under a FR1 schedule for 2 h each day. Following the acquisition of stable levels of responding across days, both naloxone pretreatment (0.25 mg/kg SC) and halving the heroin infusion dose produced increases in operant responding for heroin at each concentration. Neither ondansetron (0.01-1 mg/kg SC) nor MDL72222 (0.1-3 mg/kg SC) pretreatment influenced heroin self-administration. Chronic treatment (5 day) of ondansetron (0.01-0.1 mg/kg) was similarly ineffective. However, dexfenfluramine (0.5-2.5 mg/kg IP) consistently reduced heroin self-administration at doses producing only modest decreases in food responding. These findings are in contrast to place conditioning studies, which show that 5-HT3 antagonists but not indirect 5-HT agonists block a morphine-induced place preference. Reasons for such discrepancies remain to be determined.