There has been little investigation of the effect of neuroleptic medication on the structure of symptoms in schizophrenia. In this study, 135 male schizophrenic patients were rated with the Brief Psychiatric Rating Scale (BPRS) after 4 weeks of treatment with typical neuroleptic medication and after 2 weeks free of neuroleptics, with the order of assessment varying across patients. Confirmatory factor analyses (CFA) found that there were no differences in symptom structure across medication status and no differences in the structure of symptoms in treatment responders and nonresponders. The typical 5-factor BPRS model fit the data poorly and the fit improved considerably through deletion of items measuring symptoms not associated with schizophrenia, suggesting that some of the symptoms that contribute to a total BPRS score may be adding primarily error variance. Although the sample size in this study is limited, the results suggest that using total BPRS scores to measure severity of schizophrenic symptoms should be reconsidered.