A prospective study was conducted to assess the ability of the visually analyzed electroencephalogram (VEEG), the quantitative EEG (QEEG) and the Glasgow Coma Scale (GCS) to discriminate between patients with viral and nonviral meningitis. The 55 subjects, aged 14-75 years, fell into one of the following categories: viral (n = 12), bacterial (n = 19), tuberculous (n = 16) or cryptococcal (n = 8) meningitis. EEG recordings and Glasgow Coma Scale (GCS) scores were obtained within 48 hours of admission to hospital. The sensitivity of the VEEG and QEEG for the prediction of patients with nonviral meningitis (true positives in this context) attained reasonably high values of 70% and 80%, respectively. In contrast, the sensitivity of the GCS was only 38%. Each of the three tests achieved high degrees of consistency in this regard with positive predictive values of 94% or better. The specificity for each of the three tests was high, 100% for the VEEG and the GCS and 82% for the QEEG indicating a high probability for the correct prediction of viral meningitis (true negatives). The consistency of this prediction was, however, poor due to negative predictive values of only 53% for the QEEG, 48% for the VEEG and 32% for the GCS. The QEEG results did not reveal any obvious advantages over the VEEG. Rather the assessment of the occurrence of particular VEEG abnormalities showed that patients with delta abnormalities had a very high probability of nonviral meningitis. At the other end of the spectrum, all normal VEEGs occurred in viral meningitis. In important respects the predictive ability of the EEG was superior to that of the GCS. While there was statistically significant agreement between the VEEG and GCS, the degree of agreement was poor. This study indicates that the EEG is a valuable and probably underestimated test in the acute phase of meningitis and provides complementary information to the GCS.