ROSC is possible. Seventeen dogs were lightly anesthetized for preparation. After the dogs breathed air, the tracheal tube was clamped (simulating laryngospasm) and the dogs were immersed in ice water for 20 to 120 min. Cardiac arrest occurred at 6 to 11 rain. ECG complexes continued. At the end of submersion, the dogs were removed from the water and CPB was started using venoarterial pumping via oxygenator primed with 400 mL Dextran, 40 mL Ringer's, and heparin. Initial CPB flow of 10 mL/kg/ rain was increased as venous return allowed, to achieve normotension. The heat exchanger was kept 10 C above rectal temperature (T}. ROSC was attempted with rectal T at 32 C, using epinephrine, NaHCO3, and countershocks as needed. Life support was continued for 2 h (n = 8) or 24 h (n = 9). At start of CPB, rectal T was 34 C after 20 min and 11 C after 90 min of submersion, while cerebral T was 26 C after 20 min and 7 C after 120 min submersion. Following CPB of 0.5 to 3 h to achieve rectal T of 32 C, ROSC was successful after up tO 90 min of submersion. Limiting factor for ROSC after 90 and 120 rain submersion was clotting in large vessels. One survivor after 90 min submersion recovered spontaneous breathing and EEG activity, and so did others with shorter submersion times. CPB decreased hematocrit to 16% to 25%. There was capillary leakage. A supplementary study in rats with asphyxial arrest showed large vessel clot: ting at 20 min of normothermic arrest and at 40 min of hypothermic arrest. CPB is capable of ROSC after up to 90 min of cold water submersion. Crude signs of early cerebral recovery appear promising. CPB for outcome studies is indicated.