Depressed patients randomized to psychotherapy were compared with those who had been chosen for psychotherapy in a treatment algorithm, including addition of an antidepressant in case of early nonresponse. There were no differences between randomized and by-preference patients at baseline in adherence and outcome. About half of the early nonresponders refused the additional medication. However, no clear effect of medication addition on ultimate outcome could be demonstrated. In total, 37% of the patients achieved remission. The study suggested that randomization of patients does not induce a great influence on outcome. It might be warranted to continue an initially ineffective psychotherapy for depression, because a considerable number of patients do have a pattern of delayed response.