OBJECTIVE: To compare the cost-effectiveness of generic psychological therapy (counseling) with routinely prescribed antidepressant drugs in a naturalistic general practice setting for a follow-up period of 12 months. METHODS: Economic analysis alongside a randomized clinical trial with patient preference arm. Comparison of depression-related health service costs at 12 months. Cost-effectiveness analysis of bootstrapped trial data using net monetary benefits and acceptability curves. RESULTS: No significant difference between the mean observed costs of patients randomized to antidepressants or to counseling (342 pounds sterling vs 302 pounds sterling , p = .56 [t test]). If decision makers are not willing to pay more for additional benefits (value placed on extra patient with good outcome, denoted by K, is zero), then we find little difference between the treatment modalities in terms of cost-effectiveness. If decision makers do place value on additional benefit (K > 0 pounds sterling), then the antidepressant group becomes more likely to be cost-effective. This likelihood is in excess of 90% where decision makers are prepared to pay an additional 2,000 pounds sterling or more per additional patient with a good global outcome. The mean values for incremental net monetary benefits (INMB) from antidepressants are substantial for higher values of K (INMB = 406 pounds sterling when K = 2,500 pounds sterling). CONCLUSION: For a small proportion of patients, the counseling intervention (as specified in this trial) is a dominant cost-effective strategy. For a larger proportion of patients, the antidepressant intervention (as specified in this trial) is the dominant cost-effective strategy. For the remaining group of patients, cost-effectiveness depends on the value of K. Since we cannot observe K, acceptability curves are a useful way to inform decision makers.