After normothermic cardiac arrest in dogs, we found that mild hypothermia (34 degrees C) of 1-2 h reduced brain damage, providing that hypothermia was achieved within 15 min of reperfusion. A clinically feasible rapid brain-cooling method is needed. As head-neck surface cooling alone in dogs was found to be too slow (0.1 degrees C/min), we reviewed peritoneal cooling in the Introduction and Discussion sections. PRELIMINARY EXPERIMENTS WITHOUT CARDIAC ARREST: In 5 dogs with spontaneous circulation and IPPV, 2 L of Ringer's solution at 10 degrees C were instilled into the peritoneal cavity, left for 5 min, and drained. Brain (tympanic membrane) temperature (Tty) decreased by a mean of 0.3 degrees C/min (12 min to 34 degrees C). Core (pulmonary artery) temperature (Tpa) decreased by a mean of 0.8 degrees C/min (5 min to 34 degrees C). COOLING AFTER CARDIAC ARREST: In our reproducible dog model of normothermic ventricular fibrillation cardiac arrest of 11 min (no flow), brief low-flow normothermic cardiopulmonary bypass (CPB) was used for reperfusion and restoration of spontaneous circulation (ROSC) within 2 min. In 24 dogs, mild hypothermia was induced by head-neck surface cooling with ice bags, starting with reperfusion, plus peritoneal lavage as above, starting with ROSC. All 24 dogs were resuscitated. Initial head-neck surface cooling alone over 2 min decreased Tty by only 0.15 degrees C/min. Subsequent additional peritoneal lavage decreased Tty by a mean of 0.3 degrees C/min (11 min to 34 degrees C); and Tpa 0.6 degrees C/min (7 min to 34 degrees C). There were no significant physiologic effects. We conclude that peritoneal instillation of cold Ringer's solution is more rapidly effective than other non-intravascular cooling methods reported previously. Peritoneal cooling should be tried in patients during CPR.