OBJECTIVE: This study examined the pharmacologic, clinical, and demographic factors associated with switching antidepressants during the first three months of outpatient treatment for episodes of depression. METHODS: A cohort analysis of outpatients aged 18-75 and treated for a depressive disorder (N=56,521) was performed with PharMetrics administrative data from 2001-2006. Patients commencing antidepressant treatment who continued to receive the initial antidepressant or a second antidepressant for > or = 72 of the first 90 days were selected. Antidepressant switching was defined as prescription of a second antidepressant within 90 days of the first antidepressant prescription and continuation of the second antidepressant for > or = 15 days after termination of the first antidepressant. RESULTS: Overall, 8.6% of patients switched antidepressants during the first 90 days of treatment. The highest rates of switching were among patients initiating trazodone (47.4%), tricyclic antidepressants (26.6%), and mirtazapine (17.2%); the lowest switching rates occurred after starting venlafaxine (6.5%) or sertraline (7.4%). Antidepressant switching was significantly related to recent emergency mental health treatment (adjusted odds ratio [AOR]=1.7, 99% confidence interval [CI]=1.3-2.2); treatment of major depressive disorder versus other depressive disorders (AOR=1.4, CI=1.3-1.5), especially severe major depressive episodes (AOR=1.6, CI=1.4-1.9); and inversely related to moderately high versus low initial antidepressant dose (AOR=.7, CI=.6-.8). Several other patient characteristics were significant but less powerful predictors of antidepressant switching. CONCLUSIONS: Among adults with depression starting antidepressant therapy, medication switching commonly occurs during the first three months of treatment. Greater clinical severity and low initial dosing may increase the risk of switching antidepressants.