Steroids remain the initial therapy for acute graft-vs.-host disease (AGVHD). Strategies to improve response and minimize steroid exposure are needed. We report results of a randomized, adaptive, Bayesian-designed, phase II trial of prednisone with or without extracorporeal photopheresis (ECP) as an initial therapy for patients with newly diagnosed AGVHD. The primary endpoint was success at day 56 defined as: alive, in remission, achieving AGVHD response without additional therapy, and on <1 mg/kg at day 28 and <0.5 mg/kg on day 56 of steroids. Eighty-one patients were randomized to the ECP arm (n = 51) or steroids alone (n = 30). Median age was 54 years (range: 17-75); 90% had grade II AGVHD and 10% had grades III and IV AGVHD, with skin (85%), upper (22%)/lower (22%) gastrointestinal, and liver (10%) involvement. The ECP arm had a higher probability of success (0.815) and exceeded the predefined threshold for determining the investigational arm promising. ECP was potentially more beneficial than steroids-alone in skin-only AGVHD (response rate: 72% vs. 57%, respectively) than for visceral-organ AGVHD (47% vs. 43%, respectively). The addition of ECP to steroids may result in higher GVHD response as initial therapy for AGVHD, especially for patients with skin-only involvement.