The objective was to examine effects of risperidone on hostility and compare these effects with those of haloperidol. On the basis of risperidone's pharmacologic profile, we hypothesized that risperidone has a selective effect on hostility and that this effect is greater than that of haloperidol. The data were obtained in a multicenter clinical trial of risperidone under placebo-controlled, double-blind conditions (duration, 9 weeks). The patients were 139 patients with the diagnosis of DSM-III-R schizophrenia. Hostility was measured by the "hostility" item of the Positive and Negative Syndrome Scale. Change in hostility served as a dependent variable in the analyses. Change in "psychosis" was applied as a covariate; it helped us examine changes in hostility that were unrelated to change in psychosis (selective effect). Risperidone had a greater selective effect on hostility than did haloperidol or placebo. This finding should encourage tests of risperidone as a treatment for patients who show frequent overt physical aggression.