One hundred and one subjects, with various degrees of HIV infection, were enrolled in a longitudinal study aimed at evaluating the correlation between clinical and instrumental findings in the development of HIV-related subacute encephalitis. The method used was electroencephalography coupled with computerized spectral analysis (EEG-CSA) and mapping. The findings recorded by this method were compared with those obtained by computed tomography (CT) scan and neurological examination. The EEG-CSA findings were divided into four categories according to their severity. EEG-CSA was shown to be very sensitive in detecting the first signs of a forthcoming neurological disease. Following 11 months of observation, 22 out of 40 (55%) neurologically asymptomatic individuals who, at the beginning of the study showed some EEG-CSA abnormalities, had clinical evidence of a subacute encephalitis whereas only two out of 37 (5.4%) subjects who were previously free of EEG-CSA abnormalities had some signs of neurological disease (P less than 0.001) after the same period. Of those remaining who were already symptomatic when the study started, the neurological progression of HIV infection was also monitored by EEG-CSA.