To evaluate the efficacy of the computer-processed electro-encephalogram (EEG) for determining near-awakening from anesthesia, 14 patients were monitored during emergence from either isoflurane or fentanyl anesthesia at the termination of major surgical procedures. The raw EEG was obtained using bilateral frontomastoid electrodes. The compressed spectral array was digitized and recorded on disk in 4-s epochs using a Tractor Northern "Nomad" processor. The EEG information was displayed in four formats: 1) the frequency-power spectrum from 1-20 Hz, 2) the 95% power frequency, 3) the 50% power frequency, and 4) the ratio of power in the 8-20 Hz frequency range to the power in the 1-4 Hz frequency range (delta ratio). During emergence from isoflurane, there were obvious changes in the EEG frequency-power spectrum that occurred several minutes before patients opened their eyes in response to verbal stimuli. Although no one descriptor of EEG activity could be shown to be superior in anticipating when patients would respond by opening their eyes, awakening was always presaged by an abrupt decrease in power in the 1-4 Hz frequency range; this resulted in a marked increase in the delta ratio value. During emergence from fentanyl anesthesia, however, there was no obvious change in the overall EEG frequency-power spectrum. However, the same numeric EEG descriptors that were predictive of awakening from isoflurane also occurred during emergence from fentanyl, even though they usually occurred within 1 min of awakening. It is concluded that EEG criteria for identifying when patients will awaken from anesthesia are more reliable with isoflurane than with fentanyl.