Multimodality evoked potentials (EPs), linear electroencephalograms and Glasgow Coma Scale (GCS) scores were recorded within 24 h of cardiac arrest in 62 patients who were comatose following cardiopulmonary resuscitation. The cardiac arrest had a cardiac cause in 35 patients and a non-cardiac cause in 27 patients. The Glasgow Outcome Scale (GOS) scores were established 6 months after resuscitation. The prognostic value of all the recorded variables was calculated in terms of sensitivity, specificity and accuracy. Spearman's rank test was also used for the determination of the correlation coefficients with GOS. EP recordings furnished no falsely pessimistic predictions, with a specificity of 100%. In other words, when EPs were altered, the prognosis was always poor. However, while all patients who regained consciousness had normal EPs, not all patients in whom EPs were recordable survived. The GCS score showed a higher sensitivity and correlation with GOS score than EPs, but it was associated with a high percentage of false positive results, and its specificity was only 67%. The combination of the GCS score with EPs may be a promising strategy to counterbalance the respective limits of these methods and to reduce the loss of information due to sedation and myorelaxation, which impede clinical examination but not EP results.