We characterized prescribing in Connecticut's State public mental health system to assess the feasibility of implementing an evidence-based medication algorithm. Medication records for a random sample of outpatients with diagnoses of schizophrenia spectrum disorders showed prescribing patterns similar to the entire United States. The base rate of changing antipsychotic medications was moderate. Over half of patients received decanoate medications, polypharmacy was nontrivial, and there was variability in prescribing patterns across physicians. Caucasian patients were more likely to receive an atypical antipsychotic and less likely to have a decanoate medication, and Latino patients were less likely to change medications. Because the base rate of changing medications was moderate and a considerable proportion of patients were prescribed newer antipsychotic medications, introducing a research-derived medication algorithm with newer atypical antipsychotics as first line agents may fit well with current practice. Further, implementing such an algorithm may reduce racial and ethnic disparities in prescribing patterns.