Recovery after doxacurium and pancuronium neuromuscular blockade and their acceleration by neostigmine have not been compared in children. Therefore, 60 paediatric surgical patients aged 2-10 yr (ASA 1-2) were studied. They were randomized to receive doxacurium 30 micrograms.kg-1 or pancuronium 70 micrograms.kg-1 iv during propofol, fentanyl, isoflurane and nitrous oxide anaesthesia. Electromyographic (EMG) responses of the adductor pollicis to train-of-four (TOF) stimulation of the ulnar nerve were recorded every ten seconds using a Datex NMT monitor. Six patients in each relaxant group received neostigmine (0, 5, 10, 20 or 40 micrograms.kg-1) with atropine by random allocation when first twitch height (T1) had recovered to 25% of control. Spontaneous recovery after ten minutes was similar following doxacurium (mean +/- SEM values of 45.0 +/- 3.9 vs 49.5 +/- 10.0% for T1 and 25.2 +/- 3.8 vs 14.8 +/- 3.6% for TOF ratios). Dose-responses to neostigmine were calculated from the log dose vs logit of T1 or TOF ratio after ten minutes. Neostigmine-assisted recovery was not different in the two groups, with ED70 and ED90 doses for T1 of 14.3 +/- 1.8 and 25.7 +/- 2.7 micrograms.kg-1 for doxacurium and 12.5 +/- 1.7 and 25.3 +/- 2.3 micrograms.kg-1 for pancuronium. Time to recovery of TOF ratio to 70% after neostigmine 40 micrograms.kg-1 was 2.3 +/- 1.0 and 4.2 +/- 1.7 min (P = NS) following pancuronium and doxacurium, respectively. Adjusted recovery due to neostigmine alone (spontaneous recovery subtracted from the total) required two to three times higher doses of neostigmine.(ABSTRACT TRUNCATED AT 250 WORDS)