This study examined whether sleep disturbance predicted or moderated responses to psychotherapy in participants who participated in STEP-BD, a national, multisite study that examined the effectiveness of different treatment combinations for bipolar disorder. Participants received either a brief psychosocial intervention called collaborative care (CC; n = 130) or intensive psychotherapy (IP; n = 163), with study-based pharmacotherapy. Participants (N = 243) were defined as current (past week) short sleepers (<6 hours/night), normal sleepers (6.5–8.5 hours/night), and long sleepers (≥9 hours/night), according to reported average nightly sleep duration the week before randomization. Sleep disturbances did not predict the likelihood of recovery nor time until recovery from a depressive episode. There was no difference in recovery rates between IP versus CC for normal sleepers, and medium effect sizes were observed for differences in short and long sleepers. In this study, sleep did not play a major role in predicting or moderating response to psychotherapy in bipolar disorder.