As psychiatric practice patterns evolve to take advantage of the growing list of treatments with proven efficacy, research studies with broader aims will become increasingly important. Randomized trials may need to accommodate multiple treatment options. In completely randomized designs, patients are assigned at random to one of the options, requiring that patients and clinicians find each of the options acceptable. In "clinician's choice" designs, patients are randomized to a small number of broad strategies and the choice of specific option within the broad strategy is left up to the clinician. The clinician's choice design permits some scope to patient and clinician preferences, but sacrifices the ability to make randomization-based comparisons of specific options. We describe a new approach, which we call the "equipoise stratified" design, that merges the advantages and avoids the disadvantages of the other two designs for clinical trials. The three designs are contrasted, using the National Institute of Mental Health Sequenced Treatment Alternatives to Relieve Depression trial as an example.