The Glasgow–Pittsburgh cerebral performance categories (GP-CPC) and the Glasgow Outcome Score (GOS) have been used to categorize patients according to their neurological outcome for prognostic predictors in patients after cardiac arrest (CA). We postulated that inclusion of deaths without knowing the cerebral status into the group of patients with poor outcome after CA using the GP-CPC and GOS will lead to dilution of the prognostic power of the investigated biochemical marker. The present study was conducted to verify this issue by employing a modified outcome score, which we termed as Modified Glasgow Outcome Score (MGOS). In the present study, 97 patients were enrolled in a prospective manner. Serum NSE and S100B levels were measured daily for 7 days after admission to the intensive care unit. Neurological outcome was assessed by employing the GOS and MGOS after 6 months. By employing the GOS, 46 patients were categorized into the group of patients with poor outcome and 51 patients survived with good neurological outcome. Patients who died without certified brain damage or with unknown cerebral status after CA (n = 20) were separated from patients with poor outcome in the MGOS. The magnitude of NSE (S100B) elevation in patients with poor outcome categorized by the MGOS was approximately 1.7-fold (1.5) higher as compared with patients divided by the GOS. The mean calculated sensitivities and area under the curve values of NSE and S100B predicting poor outcome classified by the MGOS were significantly higher as compared with the GOS. Conclusively, inclusion of deaths without certified brain damage or with unknown cerebral status into the group of patients with poor outcome will lead to underestimation of the prognostic power of investigated biochemical markers such as NSE and S100B. The MGOS will help to avoid this bias.