Fifty-seven patients allocated at random to either diet supported by gastroplasty or very-low-calorie diet (VLCD) were followed for 5 years. One patient was lost to follow-up 1 year after surgery. Data were analyzed by survival statistics and success was liberally defined as a maintained weight loss of at least 10 kg. The cumulative success rate declined steadily in both groups, but it remained higher in the gastroplasty group (16 versus 3 percent, the 95 percent confidence limits being 11-21 percent and 1-5 percent, respectively; P less than 0.05). At 5 years, eight of 27 gastroplasty patients (30 percent, 95 percent confidence limits 14-50 percent) had not relapsed. Of 30 VLCD patients, relapse had not occurred in five (17 percent, 6-35 percent). The weight loss of patients without relapse did not differ significantly (medians of gastroplasty and VLCD patients 18 kg and 27 kg, respectively). Complications and side-effects were more numerous and serious among the gastroplasty patients. Although in this long-term study, the horizontal gastroplasty proved to be somewhat more effective than the diet alone, the surgical risks and the unsatisfactory maintenance of weight loss speak against the use of this operation.