Previous studies have suggested that the increased duration of action of long-acting neuromuscular relaxants may make their pharmacologic antagonism more difficult and, thus, increase the likelihood of residual block. This hypothesis was tested in healthy, adult humans who received a background of isoflurane/N2O/fentanyl anaesthesia. Study subjects were paralyzed with either pancuronium (N = 8), pipecuronium (N = 8), or the longer-acting relaxant, doxacurium (N = 8). Neuromuscular function was monitored, and, using a blinded, randomized study design, the relaxants were titrated to identify the ED95 dose in each patient. Thereafter, spontaneous recovery was observed until there was 25% of baseline response to the first supramaximal twitch (T1) in a train-of-four (TOF). At this time, the block was antagonized with neostigmine 0.07 mg/kg and glycopyrrolate 0.014 mg/kg i.v., and recovery of TOF was recorded. Spontaneous recovery to 25% of the baseline T1 response occurred at 52 +/- 14 min (mean +/- SD) following administration of either pancuronium and pipecuronium, and 85 +/- 33 min following doxacurium (P < 0.05 for doxacurium versus pancuronium and pipecuronium). In doxacurium-treated patients, reversal of block with neostigmine was less predictable and less complete than with the other two relaxants. For example, the ratio of the fourth to first twitch (T4/T1) of the TOF at 10 and 15 min after reversal was significantly less with doxacurium (59 +/- 14% and 61 +/- 16%, respectively) than with either pancuronium (75 +/- 6% and 75 +/- 10%) or pipecuronium (76 +/- 9% for both).(ABSTRACT TRUNCATED AT 250 WORDS)