Since its approval by the Food and Drug Administration in September 1985, the Garren-Edwards gastric bubble has been extensively used as an adjunct to diet and behavioral modification in the treatment of exogenous obesity. In an attempt to evaluate the efficacy of the Garren-Edwards gastric bubble, a double-blind crossover study was undertaken. Ninety patients were randomized into three groups: bubble-sham, sham-bubble, and bubble-bubble in two successive 12-wk periods. Sixty-one patients completed the entire 24-wk study. All groups participated in ongoing diet and behavioral modification therapy in a free-standing obesity program, the members of which were blinded to randomization arms. All patient groups lost weight during this study. The mean cumulative weight loss in pounds at 12 wk was as follows: bubble-sham = 19, sham-bubble = 12, and bubble-bubble = 8; and at 24 wk: bubble-sham = 23, sham-bubble = 16, and bubble-bubble = 18. The mean cumulative change in body mass index (kg/m2) at 12 wk was as follows: bubble-sham = -3.1, sham-bubble = -2.3, and bubble-bubble = -2.9; and at 24 wk: bubble-sham = -3.1, sham-bubble = -3.0, and bubble-bubble = -3.3. Although weight loss occurred more consistently in patients with a Garren-Edwards gastric bubble, there were no significant differences between any of the three groups at 12 or 24 wk with respect to weight loss or change in body mass index. The major part of the weight loss noted during this study occurred during the first 12-wk period, irrespective of therapy (bubble or sham). Side effects observed during this study included gastric erosions (26%), gastric ulcers (14%), small bowel obstruction (2%), Mallory-Weiss tears (11%), and esophageal laceration (1%). We conclude that, in this study, the use of a Garren-Edwards gastric bubble did not result in significantly more weight loss than diet and behavioral modification alone in the management of exogenous obesity, and it may result in significant morbidity.