To explore the way in which drugs act to reinstate drug-seeking behavior, we studied drug-induced reinstatement of a cocaine-induced conditioned place preference (CPP). In a series of experiments, we studied the establishment, maintenance, extinction and reinstatement of a cocaine-induced CPP in a three-chamber 'unbiased' apparatus. Groups of rats were given four 20-min pairings of one chamber with cocaine (10.0 mg/kg, i.p.) and four of the other with saline on alternate days. In 15-min tests for CPP, drug-free rats were placed in the center choice chamber with access to the entire apparatus. Experiments were designed to study the expression of the CPP, the maintenance of the CPP in tests given at 2, 4, and 6 weeks after training, and the reinstatement of the CPP by cocaine (5.0 mg/kg) after extinction by 12 repeated tests in the non-drugged state, and after extinction by pairing each chamber, on alternate days, with saline on four occasions. Robust CPPs were obtained that endured for 4 weeks and were maintained for up to 6 weeks when tests were given at 2-week intervals. Both extinction procedures led to the loss of the CPP that was in turn reinstated by priming injections of cocaine. These results indicate that a cocaine-induced CPP, once developed, endures for several weeks, and is maintained by occasional testing even in the absence of additional drug experience. The fact that the CPP is easily reinstated when testing is preceded by a priming injection of cocaine suggests that drugs may induce relapse by renewing the incentive value of drug-associated cues.