OBJECTIVES: To examine the effects of initial ziprasidone dose on discontinuation rates, using the PharMetrics integrated medical and pharmacy claims data. METHODS: Patients =18 years with a diagnosis of schizophrenia and a ziprasidone claim between March 2001 and February 2003 continuously enrolled for = 6 months before and =3 months after initiation of ziprasidone were stratified by initial daily dose (= 40mg and < 80mg [low] vs. = 80mg and < 120mg [medium] vs. 120mg-160mg [high]). The 6-month risk of discontinuation was examined using Cox proportional hazards models controlling for gender, psychiatric comorbidities, and pre-ziprasidone utilization of antipsychotics (atypical, conventional, none). RESULTS: Mean age of the sample (n = 1058) was 38 years; 42% were male. The 6-month risk of discontinuation was significantly greater in patients with a low vs high initial dose (HR = 1.357, 95% CI = 1.070, 1.721; p = 0.012) and trended towards significance when comparing a medium vs high initial dose (HR = 1.163, 95% CI = 0.905, 1.494; p = 0.237). The largest difference in discontinuation rates between dose groups occurred after the first prescription. CONCLUSIONS: Patients initiating ziprasidone therapy with an initial dose of at least 120mg/day had better medication adherence compared with those initiating at lower doses. This may reflect improved efficacy at daily doses =120mg.