The well-known amplitude reduction of the P300 appears to be unaffected by the treatment with classical antipsychotics in schizophrenia, whereas the effects of atypical neuroleptics on this event-related potential are less understood. The study of these changes could help in deciding whether the P300 amplitude reduction in schizophrenia is a trait or state marker of that illness and in better describing the effect of atypical antipsychotics on altered cognitive functions. We present a prospective longitudinal study of P300 amplitude and latency before and after 6 months' treatment with olanzapine in 11 patients with schizophrenia. A healthy control group (n = 30) was also studied. Overall, no significant changes, either in amplitude or in latency as measured at Pz and Fz electrodes, were found when comparing the pre- and postolanzapine conditions, despite the overall improvement in positive and negative symptoms. Nevertheless a direct specific association was observed between a P300 amplitude increase with olanzapine and the improvement in negative symptoms. These data would suggest that P300 amplitude reduction in schizophrenia may be relatively independent from clinical state and treatment, thus constituting a trait marker of schizophrenia. Our data also suggest that, in addition to this, some further changes in P300 amplitude might depend on the clinical state of the patients.