OBJECTIVES: To evaluate use of an optional antipsychotic drug algorithm for treating inpatients with schizophrenia or schizoaffective disorder at a state psychiatric hospital. METHODS: Clinical outcomes were compared in patients whose treatment followed a specific antipsychotic drug algorithm versus those whose did not. First step oral antipsychotic options in the algorithm were risperidone and ziprasidone. Documentation of a clinical rationale for use of a non-preferred drug was acceptable for deviating from preferred choices. Antipsychotic polytherapy was the least preferred treatment. Steps for using injectable and non-preferred drugs were also specified. Primary and secondary outcomes were length of hospitalization and patient achievement of “much improved” or “very much improved”, defined by CGI-S score, respectively. Prescribers reviewed patient record documentation to compare patients who were adherent vs non-adherent to the algorithm. RESULTS: The total cohort was 401 patients (263 algorithm adherent and 138 non-adherent). Three algorithm adherent patients were dropped due to a CGI-S score of 7 therefore, 260 were used in the analysis. Sixty-seven percent were male. The mean age was 39. The median number of past hospitalizations was 2. The modal rating of severity on the Clinical Global Impression - Severity was 5, markedly ill. There were no significant differences between groups on gender, number of past hospitalizations and severity of illness. No significant between group differences were observed for mean length of stay (adherent 49 days, non-adherent 45 days), p = 0.12, least squares means (adjusted for CGI-S, gender and exacerbations) or time to improvement, p = 0.31, log-rank test. CONCLUSIONS: Use of an optional algorithm for inpatients, designed to improve cost efficiency without denying access to non-preferred medications, did not prolong length of stay or delay time to desired improvement.