Objective To evaluate the accuracy of bispectral index (BIS) for monitoring the depth of sevoflurane anesthesia in children of different ages.Methods One hundred and five children (0-14 yr) undergoing abdominal surgery were divided into 3 age groups (n=35 each): group Ⅰ 4-14 yr (older children); group Ⅱ 1-3 yr (young children) and group Ⅲ < 1 yr (infants). All of the patients were anesthetized with sevoflurane combined with epidural or caudal blockade. Anesthesia was induced with 5 % sevoflurane in O_2. After tracheal intubation, the patients were mechanically ventilated. P_(ET) CO_2 was maintained at 35-45 mm Hg. The end-tidal sevoflurane concentration (C_(ET)Sev) Was maintained at 3.2%, 2.6%, 1.9% and 1.4% in infants < 6 months andat 2.5%, 2.0%, 1.5% and 1.0% in children > 6 months. Each C_(ET)Sev was maintained for at least 15 min. MAP, HR, P_(ET)CO_2 and BIS were monitored and recorded before anesthesia (T_1, baseline), at loss of consciousness (T_2), immediately after tracheal intuhation (T_3), at the 4 C_(ET)Sev (T_(4-7)), recovery of pharyngeal reflex (T_8), extubation (T_9) and emergence of anesthesia (T_10). Results BIS values were significantly higher in children of 1-3 yr (group Ⅱ) than in children of 3-14 yr (group Ⅰ) at T_(4,6,7). BIS values were significantly lower in infants (group Ⅲ) than in children of 3-14 yr (group Ⅰ) at T_(5-9). BIS values were significantly lower in infants (group Ⅲ) than in children of 1-3 yr (groupⅡ) at T_(3-10). There was significant negative correlation between BIS and C_(ET)Sev in all 3groups(γ=-0.768,-0.709,-0.357).Conclusion BIS can accurately reflect the depth of sevoflurane anesthesia in children (> 1 yr). BIS should be interpreted cautiously in infants (< 1 yr).