In spite of the widespread use of the Garren-Edwards gastric bubble as an adjuvant device in weight reduction, its efficacy has not been established. Therefore, our purpose was to conduct a randomized, double-blind, crossover study of this device in the management of exogenous obesity. The study group consisted of 23 patients, 21 women and 2 men, ranging in age from 21 to 53 yr. Patients were 25%-111% above their ideal body weight. They were studied for 24 wk, consisting of two separate 12-wk evaluation periods. Patients were randomly assigned either to receive the gastric bubble or to have a sham procedure. After the first 12-wk evaluation period, the gastric bubble and sham were administered in crossover fashion, so that those who had received the gastric bubble initially received the sham later and vice versa. The study coordinator remained blind to the kind of treatment, weighed each patient biweekly, enforced dietary counseling, and provided behavior modification. Those who had passed or were found to have a deflated bubble at the end of the treatment period were excluded from the study. Mean weight reduction in the two evaluation periods did not differ significantly. Patients lost 5.4 +/- 1.7 kg (mean +/- SE) during the gastric bubble period and 5.20 +/- 0.8 kg during the sham period. The order of administration of the gastric bubble and sham did not significantly affect the result. The time-course of mean biweekly values, however, revealed that with the gastric bubble, weight loss was significantly greater only during first (p less than 0.005) and second (p less than 0.025) 2-wk evaluation periods. This difference, however, disappeared after the initial 4 wk of treatment. These observations suggest that although gastric bubble implantation reduced weight significantly more than the sham procedure initially, the mean weight loss during 12 wk of evaluation was not different between the two periods. In our opinion, the gastric bubble is of no value as an adjuvant device in weight reduction.