359were "failures" and more rthan one-third actually regained all the weight lost. Moreover, though most of the "successes" or "modified successes" continued to maintain a steady weight the incidence of failure increased with the length of followup. This was especially true during the first 12 months, and follow-up reports for such short periods are of limited value.The results for the male and female patients appeared to be similar and suggest that those most overweight are least likely to succeed; however, the initial percentage excess weight was of no clear predictive value. We failed to substantiate the previous impression that the ability to reduce to within 25% in excess of the ideal weight is associated with increased prospects of long-,term sucoess. Long-term success, however, is closely correlated with regular attendance at the follow-up clinic, and only two patients were known to have kept their weight down despite a lengthy lapse of follow-up. Some of the patients who regained weight rapidly admitted that though superficially their eating habits had been reformed they continued to eat excessively but in secret.All the patien;ts were "failures" before admission and the primary reason for selection was their inability to respond satisfactorily ito conventional treatment despite regular supervision at an obesity clinic for at least a year. All were suffering either physically or psychologically from their morbid obesity. Thus for the small proporton of successes a mean weight loss of 27 kg at a mean follow-up of 27 months represents a worthwhile achievement. Success, however, cannot be gauged solely in terms of weight reduction. Ten patients had 12 elective operations, including hiatus hernia repair, cholecystectomy, and total hip replacement. Nine married, and among the women there were six pregnancies, including one full-term delivery in a previous habitual aborter. Eleven patients held jobs which pTeviously they could not have obtained. It follows that the distinction between success and failure is necessarily arbitrary, and many who regained weight did not regret the time spent in hospital.Therapeu-tic starvation could be better justified if patient selection were improved, but at present the long-term sequelae of the other form of radical therapy, bypass surgery, are uncertain. The current policy is to offer surgery only to selected patients who have regained weight after starvation.We wish to thank the many doctors, the nursing staff, and the other staff who were concerned in the care of these patients, and are grateful to Servier Laboratories for their very generous financial support. Rooth, G., and Carlstr6m, S. (1970