Pressures, hemoglobin, and oxygen saturations were measured and shunts and flows calculated before and after the surgical creation of an atrial septal defect in 20 patients with complete transposition of the great vessels. Eleven patients had intact ventricular septa and nine had ventricular septal defects. Of the group with intact septa, the shunt increased in nine and diminished in one; arterial oxygen saturation increased in nine and diminished in one; hemoglobin content decreased in eight and increased in two; systemic blood flow increased in six, diminished in two, and was unchanged in one; and pulmonary blood flow diminished in two. Of those with a ventricular septal defect, the shunt increased in all but one; arterial oxygen saturation increased in four, remained the same in two, and diminished in three; hemoglobin content increased in five and diminished in four; systemic blood flows increased in five and diminished in three; and pulmonary blood flows diminished in four. It was concluded that the procedure does not improve arterial oxygen saturation in any patient in whom the arterial oxygen saturation is 65% or more initially, and that this operation is primarily of benefit when there is an intact ventricular septum.