The feasibility of the compound electromyogram (EMG) was evaluated during onset and recovery from pancuronium block in the tibialis anterior muscle of ten cats. The evoked EMG area, amplitude and duration of the total response and of the major negative deflection were evaluated and compared to the mechanomyogram during 0.1 Hz and train‐of‐four (TOF) stimulation. EMG areas and amplitudes were found to be linearly and similarly related to the mechanomyogram during onset and recovery. Slopes of the regression lines ranged between 1.00–1.02 and between 1.10–1.22 during onset and recovery, respectively, with high individual correlation coefficients (>0.95). The TOF ratio of the mechanomyogram was linearly related to the EMG TOF ratio during onset and to the square root of the EMG TOF ratio during recovery, with no differences between EMG areas and amplitudes, suggesting a higher initial recovery of the TOF ratio of the mechanomyogram during recovery. EMG duration increased as the level of block increased but was unsuitable for neuromuscular monitoring. Evaluation of the agreement between the two methods showed that the EMG may be from 15% below to 10% above the mechanomyogram during onset and from 40% below to 45% above the mechanomyogram during recovery, in spite of high correlation coefficients. In contrast, agreement between EMG parameters was found to be high. In conclusion, EMG is more reliable than the mechanomyogram for evaluation of neuromuscular transmission in the cat. EMG amplitudes and areas both reflect the degree of neuromuscular blockade equally well