BALB/c mice were unilaterally implanted with a guide cannula, the tip of which was positioned 1.5 mm above either the lateral hypothalamus (LH) the medial hypothalamus (MH), the mesencephalic central gray area (CG), or either the dorsal (DRF) or ventral parts (VRF) of the reticular formation. On each day of the experimental period a stainless steel injection cannula was inserted into these brain structures to compare the self-administration of two doses of morphine (5 ng or 50 ng), using a spatial discrimination task in a Y-maze. At the dose of 5 ng, LH-, MH-, CG-, and VRF-injected mice all showed a regular self-administration response. At the dose of 50 ng, a discrimination between the reinforced arm and the neutral arm of the Y-maze was observed in LH-, MH-, and VRF-injected mice. Animals of the MH group exhibited the highest level of discrimination performance. At this dose, long injection latencies (> 15 min) were recorded in the CG group, which constrained us to reduce the number of daily trials from 10 to 4. In these modified conditions, CG animals clearly self-injected the dose of 50 ng of morphine. Subcutaneous injections of naloxone (4 mg/kg) reduced the number of self-administrations of morphine at each of the four responding structures. Marked signs of physical dependence (escape attempts) were observed in the four groups but with a higher frequency in CG and MH animals. When the injections of naloxone were suspended, a regular self-administration reappeared.(ABSTRACT TRUNCATED AT 250 WORDS)