The performance of the cerebral state index (CSI) in reflecting different levels of isoflurane anaesthesia was evaluated in ten cats subjected to four end-tidal isoflurane concentrations (EtIso), each maintained for 15 minutes (0.8%, 1.2%, 1.6%, or 2.0% EtIso). The CSI, hemodynamic data, ocular reflexes, and eye position were recorded for each EtIso concentration. Pharmacodynamic analysis of CSI with EtIso was performed, as well as prediction probability analysis with a clinical scale based on the eye reflexes. The CSI values showed great variability. Between all parameters, burst suppression ratio showed the better fitting with the sigmoidal concentration-effect model (R (2) = 0.93) followed by CSI (R (2) = 0.82) and electromyographic activity (R (2) = 0.79). EtIso was the variable with better prediction of the clinical scale of anaesthesia (prediction probability value of 0.94). Although the CSI values decrease with increasing isoflurane concentrations, the huge variability in CSI values may be a strong limitation for its use in cats and it seems to be no better than EtIso as a predictor of clinical signs.