Facial electromyographic activity and neuromuscular block could influence bispectral index (BIS) depth of anesthesia monitoring. In this study we examined, in 30 patients undergoing general surgical procedures, the effect of different stages of neuromuscular block on BIS monitoring and compared the conventional A-2000 BIS trade mark (BIS(3.4)) with the new BIS-XP trade mark (BIS(XP)). At deep surgical anesthesia BIS(3.4) of approximately 40, under a propofol 3.61 microg/mL target-controlled infusion and a 0.15-0.3 microg. kg(-1). min(-1) remifentanil infusion, mivacurium 0.15 mg/kg was administered. The onset of neuromuscular block triggered a brief transient odd divergence in response that manifested as a BIS(3.4) increase from 43 +/- 4 to 49 +/- 7 (P = 0.007) and a BIS(XP) decline from 41 +/- 3 to 35 +/- 3 (P = 0.003) at 1 +/- 0.2 min. Then, 2.5 +/- 1 min after mivacurium administration, both monitors returned to baseline values of 43 +/- 5 and 40 +/- 4, respectively. After that, BIS(3.4) and BIS(XP) did not significantly change during complete neuromuscular block or during various levels of neuromuscular recovery. At all phases, BIS(XP) was significantly lower than BIS(3.4). Our study indicated that the BIS(3.4)/BIS(XP) bias and the wide limits of agreement do not allow values given by the two monitors to be used interchangeably.