The influence of treatment preferences on the development of the therapeutic alliance was investigated. Seventy-five patients were followed while participating in a randomized controlled trial comparing supportive-expressive psychotherapy with sertraline or pill placebo in the treatment of major depressive disorder. Therapeutic alliance was assessed before treatment and at the 3rd, 5th, and 9th weeks of treatment. Among patients initially preferring psychotherapy, those receiving psychotherapy experienced increases in their alliance over time, whereas those receiving active medication or placebo experienced decreases. Among patients preferring pharmacotherapy, there were no differences in alliance development whether they received psychotherapy, active medication, or placebo. These relations were observed even when controlling for symptom severity. Thus, the congruence of patients' treatment preference and the treatment that they ultimately received influenced the development of the therapeutic alliance. Because alliance is a robust predictor of outcome, treatment preferences may need to be carefully considered in randomized controlled trial settings.