BACKGROUND: Adequate depth of anesthesia requires a sufficient amount of the agent to secure unconsciousness and other components of anesthesia as needed for that particular surgical procedure, without jeopardizing vital organ functions. To evaluate the relationship of depth of anesthesia to EEG, we studied the effects of increasing minimum alveolar concentration (MAC) of isoflurane (arousal, 1, 1.3, 1.5 MAC) on power spectral analysis of the EEG. METHODS: To determine 1 MAC, we studied sixty patients undergoing general anesthesia who were randomly allocated to receive isoflurane at several predetermined end-tidal concentration. A minimum of 15 min was allowed between induction and skin incision to allow steady state condition. Patients were observed for gross purposeful movement for 60 seconds after incision. The MAC was calculated using maximum likelihood solution to a logistic regression model. Another forty patients were randomly allocated to have their EEGs recorded. General anesthesia was induced with oxygen and isoflurane only. After loss of consciousness, succinylcholine 1.5 mg/kg was given and intubation followed. The EEG was recorded awake and after 15 min at steady state conditions of 1, 1.3 and 1.5 MAC isoflurane had been achieved. Spectral edge frequency 95% (SEF95), median spectral frequency (MSF), total power (TP) and relative power in the delta, theta, alpha and beta band were calculated. RESULTS: The MAC of isoflurane was 1.21 vol% (20 - 40 years) and 1.09 vol% (40 - 60 years). The distribution of spectral EEG indices of the EEGs were established and compared. The threshold value of SEF95 14 Hz to differentiate between arousal and 1.3 and 1.5 MAC had a sensitivity of 60.5% (1.3 MAC), 71% (1.5 MAC) and specificity of 74.4% (1.3 and 1.5 MAC) and that of MSF 5 Hz had a sensitivity of 71% (1.3 MAC), 81.5% (1.5 MAC) and specificity of 48% (1.3 MAC), 48.8% (1.5 MAC). CONCLUSIONS: With regard to the dose-related decrease in SEF95 and MSF under increasing end- expiratory concentrations of isoflurane as described in the present study, future studies may have todetermine whether EEG feedback control of volatile anesthetic administration may be used successfully. It seems that if neglected parts by MSF and SEF95, which are really true values are considered in the future studies, those would increase the sensitivity and specificity of EEG could be used as tool for determining depth of anesthesia.