Twenty-seven patients (3 to 22 years) underwent the Fontan operation. Seventeen had tricuspid atresia and 10 had other complex lesions. Sixteen patients had a Glenn shunt (12 prior to, three simultaneous with, and one following the Fontan operation). One early death (3.7%) occurred in a 4 1/2-year-old child with tricuspid atresia II-C, previous pulmonary artery banding, and a closing ventricular septal defect (80 mm Hg gradient). There were two late deaths (7.4%) from Candida sepsis, after 4 and 6 months, respectively. The 24 patients who survived the Fontan operation had postoperative hospital stays of 6 to 90 days (average 18). Patients with tricuspid atresia and an established Glenn shunt (nine patients, Group I) had postoperative hospital stays of 7 to 19 days (average 9.5), and none had significant pleural or pericardial effusions. Patients with tricuspid atresia without a Glenn shunt (seven patients, Group II) had postoperative hospital stays of 6 to 60 days (average 17.5), with three having significant effusions. Of the patients with other complex lesions, all without an established Glenn shunt, five had significant effusions. Four additional major complications (two tricuspid patch disruptions with ineffective pulmonary blood flow and two complete occlusions of a valved conduit) were encountered in which the Glenn shunt proved lifesaving. We believe that an established Glenn shunt played a major role in attaining minimal postoperative hemodynamic instability, effusions, renal failure, and mortality in our patients. The Glenn shunt should be considered in patients who are less than ideal candidates for the Fontan operation.