The aim of this study was to determine if the severity of paranoid/belligerence was a major determinant of neuroleptic dosage in newly admitted patients with acute or exacerbated schizophrenia. Two clinical psychiatrists, who had no clinical responsibility for drug treatment and were blind to neuroleptic dosage regimens, jointly interviewed 155 patients who were cooperative enough to be carefully interviewed with the Structured Clinical Interview for the DSM-III-R, Positive and Negative Syndrome Scale (PANSS) Edition. The large majority of the patients were receiving moderate dosages of neuroleptics (mean peak dosage: 500 mg/day of chlorpromazine equivalents). There was a positive correlation between the score on the PANSS paranoid/belligerence cluster and the daily dosage of neuroleptic treatment. Splitting the sample by gender, the correlations remained highly significant. In a multivariate analysis controlling for the effects of other clinical variables, paranoid/belligerence and gender emerged as significant predictors of neuroleptic dosage. Clinicians prescribed lower doses of neuroleptics for female patients and higher doses for patients with higher ratings on the PANSS paranoid/belligerence cluster. These findings suggest that clinicians' strategy of increasing neuroleptic dosage at the manifestation of hostility is not limited to assaultive or uncooperative schizophrenic patients who are on very high dosages of neuroleptics.