Objective To compare the accuracy of bispectral index (BIS) and cerebral state index (CSI) used to measure depth of sedation during target-controlled infusion (TCI) of propofol. Methods After obtaining written informed consent we studied 20 ASA Ⅰ-Ⅱ patients aged 25-40 years undergoing elective operation under general anesthesia. The patients were premedicated with intramuscular atropine 0.5 mg. The electrodes of BIS and CSI were placed according to the instruction manuals before induction of anesthesia. Anesthesia was induced with TCI of propofol. The target effect-site concentration was set initially at 0.5 ?g?ml-1 followed by increments of 0.5 ?g?ml-1 every 5 min until 5 min after the patients lost consciousness and did not respond to pain stimulation (OAA/S= 0) . BIS and CSI were continuously monitored and their values recorded every 2-6 seconds. OAA/S score (5 = alert, 1 = loss of consciousness) was recorded every 20 seconds. Spearman correlation coefficient between OAA/S score and BIS and CSI and their prediction probabilities (Pk) were calculated. BIS05, BIS50, BIS95 and CSI05, CSI50, CSI95 at loss of consciousness (LOC) (OAA/S = 1) were also calculated.Results CSI arid BIS correlated well with sedation depth. There was no significant difference in the prediction probability between CSI and BIS. BIS05 and CSI05 were 79.1 and 74.9; BIS50 and CSI50 67.5 and 65.9; BIS95 and CSI95 55.9 and 56.8 respectively at LOC. Conclusion During TCI of propofol both CSI and BIS can be used to measure sedation depth fairly accurately. CSI appears to be more accurate then BIS in predicting both loss of verbal contact and LOC.