A multicenter survey evaluated the clinical presentation, treatment, and outcome of accidental hypothermia. Data were collected from 13 emergency departments, with 401 of the 428 cases presenting during a two-year study period. Core temperatures ranged from 35 C to 15.6 C (mean, 30.57 C +/- 3.53) with 272 cases (63.6%) less than or equal to 32.2 C. There were no significant differences by age in presenting temperature, rewarming strategies, or mortality. The first hour rewarming rate was significantly (P less than .05) faster in the population less than or equal to 59 years (1.08 +/- 1.39 C/hr) than in those greater than or equal to 60 years (0.75 +/- 1.16 C/hr). Male core temperatures averaged 30.27 +/- 3.44 C versus female temperatures of 31.1 +/- 3.61 C. There were no clinically significant differences in male (N = 296) versus female (N = 132) profiles. High ethanol levels (315 to 800 mg%) did not affect outcome. Nine of 27 (33%) patients who received CPR initiated in the field survived, versus six of 14 (43%) with CPR begun in the ED. The profile of the CPR versus non-CPR population differed significantly (P less than .05) in location (outdoors), initial temperature (24.8 +/- 3.77 C vs 30.94 +/- 3.12 C), third-hour rewarming rate (2.28 +/- 1.53 C vs 1.17 +/- 1.18 C/hr), and numerous laboratory parameters. Tracheal intubation was performed without incident in 117 cases, of which 97 were less than or equal to 32.2 C. There were 73 fatalities (17.1%). Of these, 84.9% (N = 62) were less than or equal to 32.2 C. Predisposing conditions in this group included "serious" illness (30), systemic infection (28), trauma (15), immersion (ten), frostbite (seven), and overdose (two). The initial pulse, hemoglobin, and first-hour rewarming rate was lower in the deceased population, while the potassium, urea nitrogen, creatinine, and phosphorus were elevated. Excluding treatment combinations, outcome with exclusive use of a single rewarming strategy was passive external rewarming, 14 deaths below 32.2 C, 13 above; active external rewarming, six deaths below 32.2 C, two above; active core rewarming, 38 deaths below 32.2 C, none above. Refinements of the American Heart Association's CPR standards in hypothermia and a Hypothermia Survival Index are proposed.