Neuron-specific enolase (NSE) and S100B protein have been shown to be increased in cerebrospinal fluid (CSF) and serum of patients suffering from subarachnoid hemorrhage. This study was designed to evaluate the accuracy of NSE and S100B from CSF and serum for the prognosis of outcome and the detection of cerebral infarction, vasospasm and intracranial hypertension. In 55 patients with spontaneous subarachnoid hemorrhage and requiring external ventricular drainage the concentrations of NSE and S100B were determined daily from the serum and the CSF from admission until day 8. At ICU discharge patients' outcome was assessed by the Glasgow outcome scale and occurrence of cerebral infarction, vasospasm and intracranial hypertension were registered. Mean and peak values of each parameter for each patient were calculated. For accuracy assessment receiver operating characteristics were used. Bad outcome (Glasgow outcome scale 1 to 3) was found in 33 patients. Cerebral infarction, vasospasm, and intracranial hypertension were found in 31 (56%), 34 (62%), and 36 (65%) patients. Mean and peak values of NSECSF (P<0.001), S100BCSF (P<0.001), and S100Bserum (P<0.001) but not of NSEserum provided the ability to distinguish between patients with good and bad outcome. The accuracy of NSECSF and S100BCSF did not differ significantly from that of S100Bserum. NSECSF (P<0.001), S100BCSF (P<0.001), and S100Bserum (P<0.001) allowed the detection of cerebral infarction and intracranial hypertension. Cerebral vasospasm was detected by none of the parameters. In conclusion, NSECSF, S100BCSF, and S100Bserum provide similar prognostic values for outcome, intracranial hypertension and cerebral infarction. Significantly lower accuracy was found for NSEserum.