Methodology for assessment of depth of anaesthesia based on analysis of the electroencephalogram (EEG) is controversial. Techniques range from display of single measures, for example median value of the frequency spectrum, to dedicated pattern recognition systems based on measures of several EEG features. We have compared the performance of four techniques using tape-recorded data from 23 patients anaesthetized with either halothane or isoflurane using standardized regimens. The techniques were: (1) median frequency, (2) spectral edge frequency, (3) the cerebral function analysing monitor (CFAM1) and (4) a depth of anaesthesia monitor based on EEG pattern recognition (ADAM). Dose-response curves are presented for stepwise increases in stable end-tidal concentrations of each agent. Results indicated considerable inter-patient variability and the limitations of single EEG measures, particularly with deeper anaesthesia producing a burst suppression pattern in the EEG. Pattern recognition techniques reduced these difficulties and appeared to be promising over a wide range of anaesthetic levels.