The effect of phencyclidine (PCP), a non-competitive N-methyl- d-aspartate (NMDA) receptor antagonist, was examined in the water maze, a spatial learning and memory task dependent on hippocampal functions. Male adult C57Bl/6J mice received daily (s.c.) injections of either saline or PCP (0.25–4.0 mg/kg) for 12 days. During the last 5 days, the injections were followed by water maze training. Repeated PCP treatments disrupted spatial learning and memory in the 0.5–4.0 mg/kg dose range. Severe sensorimotor disturbances, observed at the 2.0 and 4.0 mg/kg doses of PCP, precluded further swim maze testing. The 0.5 mg/kg but not the 1.0 mg/kg dose of PCP impaired spatial learning and memory without any apparent sensorimotor deficits. PCP, at 1.0 mg/kg, produced impairment in non-spatial learning in the swim maze task and motor disturbances in the rotarod test. Repeated daily treatment with either the “atypical” antipsychotic drug clozapine (0.5 mg/kg i.p.) or the “typical” antipsychotic drug haloperidol (0.05 mg/kg i.p.) failed to influence spatial performances. The spatial impairment caused by the 0.5 mg/kg dose of PCP was blocked by concomitant treatment with clozapine (0.5 mg/kg), but not with haloperidol (0.05 mg/kg). The results suggest that it is possible, at low doses of PCP, to dissociate the spatial learning impairment in the water maze from the adverse behavioral effects of NMDA receptor blockade. This model may provide a basis for the analysis of the mechanisms underlying declarative memory disturbances in schizophrenia and the differences in mechanisms between typical and atypical antipsychotic drugs.