Disruption of an esophagogastric anastomosis can result in a high mortality despite aggressive treatment. The efficacy of fibrin "glue" to seal esophagogastric anastomoses was evaluated as a means of preventing this complication. A left thoracotomy was performed in 25 adult mongrel dogs. After esophagogastric resection, a standardized esophagogastrostomy was performed and eight interrupted sutures were used to completely close the posterior wall. The anterior wall was approximated with only three sutures, leaving four large holes between sutures. The dogs were then randomized into the control group (n = 14; no attempt to seal the leaks) or into the fibrin glue-treated group (n = 11). An average of 3.3 ml of glue was applied to the anterior wall of the anastomosis in the treated group. In the control group, 13 of 14 dogs (92.9%) died of anastomotic leak a median of 3 days after operation. In the fibrin glue-treated group, only four of 11 dogs (36.4%) died of anastomotic leaks (p less than 0.01). Dogs that survived were put to death at 14 days. Postmortem examination in all dogs revealed no deleterious effects or complications related to the glue. Postmortem examination of the one surviving control dog and the seven fibrin glue-treated dogs that did not die of sepsis revealed a healed anastomosis without abscess formation. We conclude that fibrin glue is effective in lessening the incidence of esophagogastric anastomotic leaks as employed in this experimental model.