OBJECTIVE: We compared factors associated with antidepressant treatment selection for patients initiating on duloxetine versus venlafaxine XR, bupropion, and SSRIs. We tested differences in subgroups with and without depression diagnoses. METHODS: Claims from five US managed care health plans were obtained for adult patients initiating on new prescriptions for select antidepressants between August 31, 2004 to December 31, 2004. Diagnostic and treatment history were established through prior claims (12 months before index medication date). RESULTS: Of the 230,738 eligible patients, 29.7% had depression diagnoses (n=68,526), 71.4% were female, mean age was 44.6 years, and 77.9% initiated on SSRIs. Using logistic regression models for the depression cohort, patients initiating on duloxetine (n=2061) versus all other initiators were associated with being older (OR 1.02), having more prior pain diagnoses (OR 1.11), depression-related diagnoses (OR 1.52), major depressive disorder recurrent episode diagnoses (OR 1.28), pain medications (OR 1.27), antidepressants (OR 1.46), and any psychotherapy (OR 1.14) (all p<.01). Duloxetine patients also were more likely to initiate therapy later in the study (OR 1.04; p<.0001) and were prescribed therapy by mental health (OR 2.32) and other specialists (OR 1.40) versus primary care. When depressive diagnoses were absent, duloxetine patients (n=2346) were more likely to be female (OR 1.21; p<.001) versus other antidepressant initiators (n=162,212). All other factors remained consistent. Trends over time are necessary to determine the robustness of results. CONCLUSIONS: In the first four months after the drug's availability, duloxetine initiators were associated with worse prior diagnostic and treatment histories. Case mix differences should be made when comparing drug cohorts.