Conventional blind randomized trials involve obtaining consent for treatment prior to randomization into control or treatment groups. When the trial is of necessity an open one, obtaining consent to treatment and then assigning the patient to a control group may lead to undesirable consequences. When consent is required prior to randomization, inability to obtain it diminishes both control and treatment numbers. Obtaining consent after randomization enhances the power of the study since refusal affects only treatment numbers thus making more efficient use of patient material. We propose to analyze the outcome of patients allocated to treatment, but refusing experimental therapy, as a separate group. Comparison of this group with the control population will permit validation of the efficacy of the randomization process and allow testing of the hypothesis that refusers of consent react the same as the control population. If no difference occurs we propose that the refuser and control groups be combined and compared with those patients who receive experimental treatment.