Fifty-seven morbidly obese patients were randomized to gastric bypass (29) or gastroplasty (28) and observed for 24 months postoperatively. Operating time and hospital stay were longer and peroperative and postoperative complications somewhat more common in the bypass than in the gastroplasty group. But weight loss at 1 year was significantly greater and failures significantly fewer after gastric bypass. Four gastroplasties were converted to bypass after 18-24 months because of failure to lose weight. Gastric bypass was judged to be much the preferable of the two operations. Dumping occurred in some patients with bypass but not after gastroplasty. Dumping was not associated with greater weight loss. Peroperatively measured pouch volume showed significant correlation with weight loss after gastroplasty, but not after bypass. Peroperative pouch volume and postoperatively measured stoma diameter were co-acting factors, which in multiple regression analysis could explain observed variance in weight loss to about 40% after gastroplasty, but to a negligible degree after gastric bypass.