We compared the performance of three newly developed anesthetic agent (AA) monitors: the Bruel & Kjaer Anesthetic Agent Monitor 1304 (BK 1304), the Datex Capnomac Ultima (ULTIMA), and the Nellcor N-2500 (N-2500). The following were investigated: the linearity and accuracy in measuring AAs, oxygen, carbon dioxide, and nitrous oxide; the linearity and accuracy during warm-up time; the effect of increasing respiratory rate on the accuracy; the consequences of a difference between monitored and delivered AA and of delivering a mixture of AAs; and, finally, the effect of water vapor and alcohol. For all three monitors we found that the accuracy in determining the respiratory and anesthetic gases was sufficient for clinical use (the N-2500 does not measure oxygen). Because of the calibration mixture supplied with the device, however, the ULTIMA recorded values that were 10 to 12% (relative) less than the AA that was present. The BK 1304 had greater accuracy at higher respiratory rates than did the other two monitors, probably favoring its use in pediatric anesthesia. The N-2500 will detect which agent (isoflurane, enflurane, or halothane) is being used, alone or in a mixture. With the two other monitors the user must define which agent is given. In some situations a difference between this and the one actually delivered can theoretically lead to an overdose of AA, with the ULTIMA up to a 14.9 minimal alveolar concentration (MAC) overdose. No interference from alcohol or water vapor in the expired air was found.(ABSTRACT TRUNCATED AT 250 WORDS)