Rats were implanted with stimulating electrodes aimed either at the medial forebrain bundle-lateral hypothalamus (MFB-LH) or the midbrain-central gray (MID-GG), and were trained to lever-press for brain self-stimulation on a fixed ratio: 15 schedule of reinforcement. The dose-dependent effects of morphine (0.1-3.0 mg/kg), naloxone (0.1-30 mg/kg), and naltrexone (0.1-30 mg/kg) were then determined during 1 h test sessions. Both naloxone and naltrexone decreased the rate of responding in the MFB-LH as well as in the MID-CG. However, decrements in response rates were produced in the MID-CG by both naloxone and naltrexone at one tenth the doses required to produce similar decrements with electrodes in the MFB-LH. Dose-dependent decreases in response rates produced morphine occurred at the same doses in the two electrode sites. At both sites, the decreases in response rates produced by the highest dose of morphine were antagonized completely by a low dose of naloxone (0.1 mg/kg). At an intermediate dose of naloxone (1.0 mg/kg), antagonism occurred in the MFB-LH but not in the MID-CG. At a high dose of naloxone (10 mg/kg), a depression in lever-pressing occurred at both sites in the morphine-treated animal indicating that the depressive action predominated over antagonism. These data explain the lack of consistency of the effects of naloxone on brain self-stimulation previously reported by different laboratories, and demonstrate that the use of partial reinforcement schedules in a rational approach to the evaluation of opioid effects on brain self-stimulation behavior.