The purpose of this experiment was to examine the effects of a nondrug alternative reinforcer and feeding conditions on the acquisition of cocaine self-administration. Rats were autoshaped to press a lever that resulted in a 0.2 mg/kg i.v. cocaine infusion. Responses on the lever were monitored during six consecutive autoshaping sessions that occurred each day. A retractable lever was inserted into the operant chamber on a random time 60 s schedule 10 times per session for six sessions that began each hour. Each day the six autoshaping sessions were followed by a 6-h cocaine self-administration session. During self-administration the lever remained extended, and each response on the lever resulted in a cocaine infusion (0.2 mg/kg). The criterion for acquisition of cocaine-reinforced behavior was met when there were 5 consecutive days during which the mean number of infusions during the 6-h self-administration session was at least 100. This procedure was repeated daily until the criterion was met or 30 days elapsed. The rats were also trained to respond on lick-operated automatic drinking devices that delivered 0.05 ml water or a glucose and saccharin solution (G + S) contingent upon each lick response. Five groups of 12-14 rats were compared. The first four groups constituted a 2 x 2 factorial design whereby either G + S or water was available in the home cage for 3 weeks before autoshaping began and G + S or water was available in the operant chamber during autoshaping. These groups were limited to 20 g food per day and all had free access to water.(ABSTRACT TRUNCATED AT 250 WORDS)