STAR*D was designed to assess competing treatments for patients with major depressive disorder who presented for care in primary and specialty clinics and who did not reach remission after a prospective trial with citalopram. Inclusion and exclusion criteria were limited so that a broad spectrum of patients would be studied. Treatments were open while assessments were blind. Equipoise randomization took into account patient and clinician preferences for blocks of competing treatments for 3 subsequent levels of treatment. Generally, patients opted to either switch or augment medication, with few agreeing to either do both or to agree to have psychotherapy as an option. About a third of patients reached remission with citalopram, a third reached remission after the second level of treatment, and then less than 15% reached remission for each of the two remaining levels. First pass analyses suggested that for those patients who reached remission or who responded without remission but who elected to continue on those treatments anyway, the relapse and recurrence rates approached 50% and greater. The overall lesson from STAR*D is that our current treatments are limited and few reach and sustain remission over the course of a year.