In a prospective randomized study of pyloroplasty versus no drainage, 200 patients (100 in each group) in whom the whole stomach was used for reconstruction following resection for esophageal carcinoma were studied. Only patients who underwent the Lewis-Tanner operation and who had a normal pylorus were included. There was no morbidity from the pyloroplasty procedure. Thirteen patients without drainage developed symptoms of gastric outlet obstruction, requiring prolonged post-operative parenteral nutrition, and reoperation was required in one patient. Four patients developed pulmonary complications associated with gastric distension, which resulted in fatal aspiration in two patients. Five patients had symptoms of outlet obstruction with eating at the time of their death. Mean and standard deviation of daily gastric aspirate was 161 +/- 88 mL in the pyloroplasty group and 233 +/- 142 mL for the control group (p = 0.23). Gastric emptying test showed mean T1/2 +/- standard deviation of 6.6 +/- 7.5 minutes in the pyloroplasty group and 24.3 +/- 31.5 minutes in the control group (p less than 0.001). More patients in the pyloroplasty group were able to tolerate a solid diet and at normal or increased amounts than were patients in the control group in the early postoperative weeks (p less than 0.01). In addition, control patients were found to have increased symptoms with meals, which were more frequent and of greater severity than symptoms in patients in the pyloroplasty group, even at 6 months after surgery (p less than 0.01). Therefore, we recommend a pyloroplasty for patients in whom the whole stomach is used for reconstruction after esophagectomy.