OBJECTIVE: To investigate long-term patterns of antidepressant treatment in patients in primary care in the UK, and to assess their healthcare resource use and disease outcomes. RESEARCH DESIGN AND METHODS: A retrospective longitudinal cohort study was conducted using the Clinical Practice Research Datalink. The study population comprised patients aged >/=18 years with depression receiving a prescription for antidepressant monotherapy between 1 January 2006 and 31 December 2011 with no antidepressants within the preceding 6 months. Recovery was defined by timing of antidepressant prescriptions (>/=6 months without treatment). Treatment lines and strategies (switching, combining, augmenting and resuming medication) were analyzed. Healthcare resource use for the different treatment strategies and periods of no therapy was assessed. RESULTS: Data from 123,662 patients (287,564 treatment lines) were analyzed. Switching and resumption of treatment were more frequent than other strategies. Recovery was highest with first-line monotherapy (45% of patients), while as a second-line strategy switching was more successful (43%) than combination or augmentation. In subsequent lines of treatment, switching was associated with successively lower rates of recovery (31% in the third line and 24% from the fourth line onwards). Similar rates were observed for resumption. Healthcare resource use was greater during antidepressant use than treatment-free periods. Augmentation was associated with the highest proportions of patients with a psychiatrist referral, psychologist referral and psychiatric hospitalization. CONCLUSIONS: This study provides extensive real-world information on the prescribing patterns and treatment outcomes for a large cohort of patients treated for depression with antidepressants in primary care. Switching is more frequently used than augmentation or combination treatment, with decreasing effectiveness across successive lines. Key limitations of the study were: (i) risk of selection bias due to the use of inclusion criteria based on depression diagnoses recorded by the practitioner; and (ii) reliance on prescribing patterns as proxies for clinical outcomes, such as recovery.