Objective. The present study is to clarify whether the bias and limits of agreement of the TOF-Guard and the mechanomyograph differ from those of two mechanomyographs on contra lateral arms. Previous studies of the bias and limits of agreement between acceleromyographical (TOF-Guard®) and mechanomyographical measurements of neuromuscular transmission did not take the error introduced by using contra lateral arms into consideration. Methods. Fifty-two women undergoing gynecological surgery were anesthetized with midazolam, fentanyl, thiopental, halothane and nitrous oxide. Neuromuscular blockade was induced and maintained with atracurium. In 32 patients, neuromuscular monitoring was performed with a Myograph 2000® on one hand and a TOF-Guard® at the other (M/T group). In 20 patients, monitoring was performed with a Myograph 2000® at both hands (M/M group). Train-of-four stimulations were applied to the ulnar nerve at the wrist in both groups. Bias and limits of agreement between the contra lateral hands in each group were calculated as proposed by Bland and Altman. Results. When the TOF ratio was 0.25, TOF ratio bias and limits of agreement in the M/T group were 0.86 and 17.58 to −15.85, respectively. Corresponding values in the M/M group were −1.75 and 12.3 to −8.8. Bias in the M/T group decreased significantly to −8.1 when TOF ratio increased to 0.70, resulting in limits of agreement of 12.1 to −28.4. The corresponding values in the M/M group were bias 2.0 and limits of agreement 10.7 to −6.7. TOF-Guard® bias for onset time and time to 5% recovery of T1 (first twitch in TOF) were −19 s and −1.5 min, respectively; both values differed significantly from zero (P < 0.05). Taken together with the changing TOF-ratio bias during recovery in the M/T group, these results indicate different onset and recovery curves for the two monitoring devices. Conclusions. Due to wide limits of agreement and different recovery courses, acceleromyographic and mechanomyographic recordings of neuromuscular transmission cannot be used interchangeably. The substantial variation between simultaneous mechanomyographical recordings of neuromuscular transmission obtained in contra lateral arms suggests that this factor should be taken into account when studying new neuromuscular monitoring techniques using the two-arm technique.