Thirty-eight patients considered to be at increased risk for a Fontan repair underwent bidirectional cavopulmonary anastomosis. Twenty-one of the 38 had concurrent pulmonary artery reconstruction. Fontan risk factors included pulmonary artery distortion, elevated pulmonary artery resistance (greater than 2 Woods units) and/or pulmonary artery pressure (mean, greater than 18 mm Hg), atrioventricular valve regurgitation, systemic ventricular dysfunction, complex venous anatomy, and subaortic obstruction. There were no deaths, either early or late. Median arterial oxygen saturation increased from 79% to 84% (p less than 0.01). Median hospital stay was 8 days. No patient had pleural effusions after 7 days. Three patients had significant surgical complications. Five patients had inadequate relief of cyanosis; three of these had venous collaterals and two had severe ventricular dysfunction; the latter two patients subsequently had strokes. One of the patients with persistent cyanosis required a systemic-to-pulmonary artery shunt. We conclude that a bidirectional cavopulmonary shunt, with pulmonary artery reconstruction when indicated, provides adequate relief of cyanosis in most patients with single-ventricle lesions who are considered to be at increased risk for a Fontan repair. Relief of ventricular volume overload and pulmonary artery distortion may improve Fontan candidacy. Also, patients with persistent cyanosis after bidirectional cavopulmonary anastomosis should be catheterized for location and occlusion of venous collaterals. Further follow-up is necessary to determine the place of bidirectional cavopulmonary anastomosis in the management of patients at increased risk for a Fontan repair.