OBJECTIVES:  To examine healthcare cost patterns prior to and following duloxetine initiation in patients with major depressive disorder (MDD), focusing on patients whose dose was initially high or was titrated to high levels. METHODS:  This was a retrospective analysis of 10,987 patients aged 18 to 64 years treated in a real-world clinical setting and enrolled in health insurance for the 6 months preceding and 12 months following initiation of duloxetine. Repeated measures and pre-post analyses were used to examine longitudinal healthcare cost trajectories before and after initiation of low- (<60 mg/day), standard- (60 mg/day), and high- (>60 mg/day) dose duloxetine. Classification and regression tree analysis was used to identify factors influencing patient heterogeneity in healthcare cost outcomes for patients whose doses were titrated to high levels. RESULTS: Low, standard, and high doses of duloxetine were initiated for 29.6%, 60.9%, and 9.5% of  patients, respectively, with 13.7% received titrated high dose within 6 months duloxetine initiation. Regardless of dose, total costs increased in the months leading to and decreased in the months following initiation of treatment. Patients whose dose was initially >60 mg/day had higher costs both prior to and throughout the course of treatment as compared to patients treated with standard-dose duloxetine. Following dose escalation to >60 mg/day, costs were higher for medication but lower for inpatient services, resulting in total cost neutrality. Titration to high-dose therapy was cost beneficial for patients with histories of a mental disorder and high prior medical costs. CONCLUSIONS: Patients with MDD had cost increases leading to and cost declines following initiation of duloxetine therapy. Patients treated with >60 mg/day had higher healthcare costs both prior to and following initiation compared to those treated with ≤60 mg/day. Increases in pharmacy costs associated with escalation to high-dose therapy were offset by reduced inpatient costs.