The effects of cocaine HCl infusions into either the nucleus accumbens (NACC) or medial prefrontal cortex (PFC) were compared on the performance of schedule-induced polydipsia (SIP) and related behaviours. Food-deprived rats were exposed to a fixed-time 60-s schedule of food delivery in daily 30-min sessions until stable levels of behaviour were obtained (14 days). Rats were then bilaterally infused with cocaine into either the NACC or PFC via chronically indwelling guide cannulae. Each subject received a sequence of five cocaine infusions (0, 12.5, 25, 50, 100 micrograms) according to a Latin Square design. For comparison, following these intracranial infusions each rat received a sequence of five IP injections of cocaine (0, 2.5, 5, 10, 20 mg/kg) also in a counterbalanced order. NACC and PFC infusions of cocaine and IP cocaine dose-dependently reduced SIP. Cocaine infusions into the NACC, but not the PFC, increased locomotor activity but the characteristic temporal profile of locomotor activity during SIP was retained. IP cocaine also increased locomotor activity in a dose-dependent manner, but the temporal profile of activity was flattened following 20 mg/kg cocaine. NACC and PFC infusions of cocaine had little effect on the total number of panel presses to gain access to the food pellets, but did slightly decrease the high rates of responding immediately prior to the pellet delivery. IP cocaine increased the total number of panel presses at the higher doses, mainly by increasing the low rates of responding. The effects of cocaine infusions into the PFC were behaviourally the most selective, as they reduced SIP without having substantial effects either on locomotor activity or panel pressing.(ABSTRACT TRUNCATED AT 250 WORDS)