The management of intrathoracic esophageal perforation with delayed diagnosis is a subject of controversy. Because of the obvious advantages of primary repair as a simple single-stage operation, this technique was preferentially used to treat 18 of 22 consecutive patients with esophageal perforation. These patients were stratified into three groups according to the time interval between perforation and repair: group A, less than 6 hours, five patients (28%); group B, 6 to 24 hours, six patients (33%); and group C, more than 24 hours, seven patients (39%). Group A patients were older (p < 0.05) and group B had fewer iatrogenic perforations (B, 17%; A, 80%; C, 57%, p < 0.1). Additional tissue was used to buttress the repair site in all three groups (A, 3/5 patients, 60%; B, 4/6 patients, 67%; C, 6/7 patients, 86%; p = not significant). In seven patients (39%), a fundic wrap was used to reinforce the site of primary repair. The outcomes of the three groups were analyzed. Group A had the lowest proportion of postoperative leaks (A, 0/4 patients, 0%; B, 4/6 patients, 67%; C, 5/6 patients, 83%; p < 0.05) and postoperative morbidity (A, 2/5 patients, 40%; B, 6/6 patients, 100%; C, 6/7 patients, 86%; p < 0.1). However the increased incidence of leak and morbidity did not lead to an increase in mortality. One death occurred in each group, with an overall mortality of 17% (A, 1/5 patients, 20%; B, 1/6 patients, 17%; C, 1/7 patients, 14%; p = not significant). We conclude that in the era of advanced intensive care capabilities, primary repair of intrathoracic esophageal perforation can be safely accomplished in most patients regardless of the time interval between perforation and operation. Leakage at the suture site is common unless primary repair is carried out without delay. Postoperative leakage, however, is usually inconsequential and does not necessarily result in an adverse outcome.