The results of this study should be interpreted with caution because of certain methodological limitations. First, the open-label design of the study may have influenced the assessment of psychopathology and side effects to an unknown degree. However, the lack of significant differences among the three groups with regard to the frequency and severity of extrapyramidal side effects and other adverse events at the end of the study suggests that these biases were probably minimal. Second, the possibility that a higher dose of risperidone could have been associated with an even lower relapse rate cannot be excluded. This is unlikely, however, because the actual relapse rate of the maintenance group with no dose reduction (7.8%) compares favorably with the results of similar studies. Third, the results are not applicable to other antipsychotic drugs. Fourth, the patients' reasons for discontinuation were not recorded. Fifth, although there are several pathways to meeting the relapse criteria of Csernansky et al. (11), the characteristics of those pathways were not recorded and analyzed. Finally, an exploration of the association between initial doses and relapse would have been of interest, particularly a determination of whether a 50% reduction of high initial therapeutic doses (6–8 mg/day) could be well tolerated and effective. However, as there were only 11 patients in the two dose-reduction groups with initial daily doses of 6–8 mg, no statistical analysis could be performed.