Aim of the study: To investigate serum levels of glial fibrillary acidic protein (GFAP) for evaluation of neurological outcome in cardiac arrest (CA) patients and compare GFAP sensitivity and specificity to that of more studied biomarkers neuron-specific enolas (NSE) and S100B. Method: A prospective observational study was performed in three hospitals in Sweden during 2008-2012. The participants were 125 CA patients treated with therapeutic hypothermia (TH) to 32-34. °C for 24. hours. Samples were collected from peripheral blood (n. =. 125) and the jugular bulb (n. =. 47) up to 108. hours post-CA. GFAP serum levels were quantified using a novel, fully automated immunochemical method. Other biomarkers investigated were NSE and S100B. Neurological outcome was assessed using the Cerebral Performance Categories scale (CPC) and dichotomized into good and poor outcome. Results: GFAP predicted poor neurological outcome with 100% specificity and 14-23% sensitivity at 24, 48 and 72. hours post-CA. The corresponding values for NSE were 27-50% sensitivity and for S100B 21-30% sensitivity when specificity was set to 100%. A logistic regression with stepwise combination of the investigated biomarkers, GFAP, did not increase the ability to predict neurological outcome. No differences were found in GFAP, NSE and S100B levels when peripheral and jugular bulb blood samples were compared. Conclusion: Serum GFAP increase in patients with poor outcome but did not show sufficient sensitivity to predict neurological outcome after CA. Both NSE and S100B were shown to be better predictors. The ability to predict neurological outcome did not increased when combining the three biomarkers.