BACKGROUND Double burst stimulation (DBS) was originally introduced for improved manual detection of residual neuromuscular blockade. Previous studies demonstrated a high correlation between mechanomyographical responses to DBS and train-of-four (TOF) stimulation during recovery from neuromuscular blockade. However, repeatability and bias analyses that are recommended when new monitoring devices are introduced into clinical practice [11] have not yet been performed. OBJECTIVE The object of the present study was to evaluate if DBS3,350/50 (3 stimuli at 50 Hz followed 0.750 sec later by 3 stimulations at 50 Hz) and TOF measurements are in so close agreement that they can be used interchangeably during spontaneous recovery from atracurium blockade. METHODS The study comprised 20 women undergoing gynaecological laparotomy and anaesthetised with fentanyl, thiopentone, halothane, and nitrous oxide. The neuromuscular blockade was induced and maintained with atracurium. The ulnar nerve was stimulated using DBS or TOF stimulation. Neuromuscular transmission was monitored mechanomyographically. Alternating sequences of 4 DBS and 6 TOF stimulations were applied during the phase of spontaneous recovery. Repeatability, bias (accuracy) and limits of agreement were calculated as proposed by Bland and Altman [11]. RESULTS The repeatability coefficients before any neuromuscular blocking agent was given were 3.4 and 7.7% for T1 and D1, respectively (P < 0.05), and 3.8 and 3.5% for TOF ratio and DBS ratio, respectively (P > 0.05). The mean difference between duplicated DBS and TOF measurements during recovery (repeatability) differed from zero due to the ongoing recovery process. It was therefore not possible to calculate the repeatability coefficients. The DBS ratio bias decreased from 6.69 to 3.51% (P < 0.05) during recovery. The limits of agreement between the DBS and TOF ratios increased from -2.07 to 15.45%, to -11.93 to 18.95% during recovery, while the limits of agreement between the DBS and TOF twitch heights increased from -5.02 to 10.68%, to -21.02 to 25.26%. CONCLUSIONS The limits of agreement between DBS and TOF responses were so wide that DBS and TOF can not be used interchangeably.