Following recommended dosage and duration protocols from the start of a treatment episode is essential to achieving optimal clinical outcomes through pharmacotherapy. For retrospective claims database analysis of antipsychotic medications, early refills and/or dose titration can complicate comparisons of an individual patient’s dose and duration pattern to recommended protocols. OBJECTIVE: The objective was to determine whether an algorithm adjusting for early refills and dose titrations in antipsychotic claims resulted in significant changes. METHODS: Antipsychotic prescriptions dispensed between October 1995 and December 1997 to 21,536 eligible subjects were extracted from a large US prescription claims database. Patients starting a new episode—initiators (n=3,244)—received their initial antipsychotic medication between October 1996 and December 1996, but received no antipsychotic agents in the preceding 365 days. Antipsychotics dispensed within 365 days after the initial agent were used to calculate mean daily dose and duration for initiators by medication. Paired t-tests assessed differences between patient-level dose and duration measures calculated with and without the algorithm. RESULTS: Use of the algorithm significantly reduced duration and significantly raised mean daily doses for initiators taking one of seven antipsychotic medications. For three agents—olanzapine, risperidone, and thioridazine—both dose and duration differences were significant at p?0.0001. Dosing and duration p-values, respectively, for the other agents were: chlorpromazine (0.0024, 0.0329), haloperidol (0.0017, 0.0002), prephenazine (0.0101, 0.0003), and trifluoperazine (0.0348, 0.0026). CONCLUSIONS: Applying the algorithm resulted in significant changes to antipsychotic dose and duration values. Therefore, before comparing patterns developed from retrospective antipsychotic claims to recommended guidelines, dose and duration must be adjusted to account for early refills and/or titration.