Of 347 victims of out-of-hospital cardiac arrest 196 (56.5%) died before and 109 (31.4%) after admission to hospital, while 42 patients (12.1%) were discharged alive. The 37 patients (10.7%) discharged without severe hypoxic brain damage were assigned to the group with "good", the remaining 310 patients to the group with "poor outcome". From results of stepwise logistic regression, a score was derived to specifically identify victims with poor prognosis (values in brackets = score points; cutpoint: score greater than 3 points): age less than or equal to 70 (0), 71-80 (1), greater than 80 (2); ECG ventricular fibrillation (0), other (1); no aspiration (0), aspiration (1); pupils round (0), not round (1); gasping (0), apnea (1); bystander resuscitation--yes (0), no (1). Evaluation of the score revealed a specificity of 100% (0.95 confidence interval: 80%-100%) and predictive value of 100% (0.95 confidence interval: 95%-100%). A predictive score for specific identification of victims with poor prognosis can contribute to decision making in out-of-hospital cardiac arrest.