OBJECTIVE: This study documented the relationship between antipsychotic drug use patterns and direct health care costs for community-dwelling patients with schizophrenia in four state Medicaid programs. METHODS: A one-year treatment period was defined for each patient (N=18,090). Logistic regression models were used to investigate the predictors of not having used any medication, delaying therapy, changes in therapy, and completing one year of continuous therapy. Regression analyses were used to estimate the impact of these patterns of antipsychotic drug use on direct health care costs. RESULTS: The average annual total direct cost for community-dwelling patients with schizophrenia was $13,650 (N=18,090). Approximately 24% of patients did not use any antipsychotic drug for at least one year and consumed significantly fewer services (-$3,200). Delaying antipsychotic therapy (18%) was associated with an increase in total cost of care of $3,936 (p = 0.0001). Patients who switched or augmented their initial therapy (41%) were also found to experience higher total cost of care ($4,019, p = 0.0001). Continuous antipsychotic therapy (18%) was associated with significant shifts in the type of care consumed (-$1,337 in hospital costs; -$652 in nursing home costs; +$1,237 in ambulatory services and +$939 in other services), leaving total cost unchanged. CONCLUSIONS: This study confirms that patients with schizophrenia treated with traditional antipsychotic medications did not display drug utilization patterns consistent with successful drug therapy. Sub-optimal drug use patterns were associated with increased direct healthcare costs.