Fifteen of 194 patients (7.7%) with tetralogy of Fallot operated upon since January 1, 1972 under a protocol of routine primary repair despite young age died in-hospital. Most deaths were from low cardiac output. Young age and smallness of size increased the risk of operation. No deaths occurred among patients older than 4 years. High hematocrit was also a risk factor. Transannular patching has an independent effect in increasing risk. The post-repair ratio of peak pressure in the right ventricle to that in the left did not exert an independent effect. To project current risks of a two-stage approach, we determined that five of 158 patients (3.2%) died in-hospital after secondary intracardiac repair after a previous Blalock-Taussig or Waterston anastomosis between 1967-1978. Using these data and those we have published on the risk of shunting, we project that except in very small babies, the risks of hospital death of a two-stage approach are not less than those of primary repair done without a transannular patch, except when body surface area is less than about 0.35 m'. When a transannular patch is used in the primary repair, the two-stage approach is projected to be safer when the child has a body surface area of about 0.48 m' or smaller.