OBJECTIVE To investigate the use of the Narcotrend electroencephalogram (EEG) monitor for the assessment of anesthetic depth in horses undergoing xylazine balanced isoflurane anesthesia. METHODS Blinded experimental study. METHODS Seven healthy warm-blooded horses, aged 10.6 ± 5.9 years, weighing 535 ± 55 kg. METHODS Horses were anesthetized for a terminal surgical trial with xylazine, thiopentone, and guaiphenesin for induction and isoflurane and xylazine continuous rate infusion for maintenance. After surgery, an EEG was recorded and processed by the Narcotrend monitor. It displays an index [Narcotrend index (NI)] between 0 and 100, which is supposed to indicate anesthetic depth. This index was recorded and correlated with eight different end tidal (ET) isoflurane concentrations between 0.8 and 2.2 vol%. In addition, anesthetic depth was numerically scored based on common clinical signs with a score of 1 (plane "too deep") to 4 (plane "too light"). After testing for normal distribution, both clinical scores and NI were correlated with different ET isoflurane concentrations using Spearman rank correlation. RESULTS Correlation of NI with ET isoflurane concentrations was poor (r s = 0.24). The NI ranged between maximal 48 and minimal 13 in the horses. The clinical scores decreased with increasing ET isoflurane concentrations (r s = 0.80). They ranged from 1 to 4 in different horses at the concentrations investigated. CONCLUSIONS In this study, the NI did not seem to be useful for assessment of anesthetic depth in horses receiving isoflurane anesthesia balanced with a xylazine constant rate infusion.