OBJECTIVES Patient adherence to pharmaceutical therapy is an important factor in the successful treatment of depression. Previous research has shown anti-depressant medication adherence associated with patient age, gender, and medical history; physician specialty; and practice patterns. Socioeconomic factors and fiscal incentives stemming from health plan benefit designs have also been hypothesized to impact adherence. In this study, we examined how patient, physician, and health plan benefit design characteristics impact compliance and persistence, two common measures of adherence. METHODS Retrospective database analysis among patients receiving at least one prescription for a selected anti-depressant (duloxetine, venlafaxine XR, escitalopram, sertraline, paroxetine, fluoxetine, citalopram, and/or bupropion) between October 1, 2004 and September 30, 2005. The study cohort included individuals, age 18-64, with a recorded depression diagnosis code within 6 months of the initial index drug prescription (n=126,889) and continuous benefit coverage pre- and post-index date (first recorded prescription date for the index drug). Compliance was measured as percentage of patients with 365-day medical possession ratio >/=80%; persistence, as days on therapy (DOT) with index medication. Predictors of compliance were assessed with a logistic regression model; persistence, with a Cox Proportional Hazards model. RESULTS Predictors of lower compliance included African-American race, index drug initiation as second or higher line of therapy or at subtherapeutic dosing levels, psychotropic polypharmacy, comorbid pain diagnoses, and co-pay level (95% confidence intervals for odds ratios all >1). Predictors of higher compliance included older than 45, more than high school education, income > $100K, and comorbid anxiety (95% confidence intervals for odds ratios all >1). Similar results were obtained when assessing persistence instead of compliance. CONCLUSIONS When possible, comparisons of adherence to alternative antidepressant medications should be adjusted for elements of benefit design, socioeconomics, patient demographics, and physician specialty and treatment patterns.