To determine the preferable reconstruction of gastrointestinal continuity after gastric bypass, we studied by endoscopic, chemical, and histologic analyses 28 randomly selected patients with a loop gastroenterostomy, a loop gastroenterostomy plus enteroenterostomy between the afferent and efferent loops, and a Roux-en-Y anastomosis. Total bile acid levels for the three groups were: 5092 +/- 1673 mumol/L, 1638 +/- 581 mumol/L, and 404 +/- 384 mumol/L, respectively. The incidence of gastritis by endoscopy was 71% in the standard loop bypass, 45% in the enteroenterostomy group, and 13% in the Roux-en-Y group. Histologic abnormalities were present in 86% of the patients who underwent standard loop bypass, in 91% of those with an additional enteroenterostomy, and in 63% of the Roux-en-Y group. There was poor correlation of symptoms and objective findings. In our study Roux-en-Y reconstruction after gastric bypass, in comparison with loop gastroenterostomy or loop gastroenterostomy with an additional enteroenterostomy, is less likely to result in bile in the stomach, endoscopic changes, and histologic abnormalities.