Residual paralysis following the use of neuromuscular blocking drugs remains a clinical problem. As part of departmental quality assurance, we examined the degree of postoperative residual curarization (PORC) following atracurium. Forty patients undergoing general anaesthesia involving atracurium were studied. Quantitative neuromuscular monitoring (mechanomyography, Myograph 2000, Biometer, Denmark) was performed by assessing the response to supramaximal train-of-four (TOF) stimulation of the ulnar nerve. Anaesthesia was provided by non-participating clinicians who were blinded to the study data. A TOF ratio </=0.7 at extubation was classified as PORC. At antagonism of neuromuscular block, 70% (28/40) of patients had a TOF ratio </=0.7, and 65% (26/40) of patients had a TOF ratio </=0.7 at extubation. Peripheral nerve stimulator use was associated with a longer interval from antagonism of block to extubation (P=0.01), but was not associated with differences in atracurium dosage or a reduction in PORC at extubation. Patients with TOF ratio </=0.7 at extubation had surgery of shorter duration [59 (SEM 6) vs 103 (9) min, P<0.001], greater doses of atracurium relative to the duration of surgery [6 (1) vs 11 (1) micro g kg(-1) min(-1), P<0.005], and shorter intervals from administration of last dose of atracurium to antagonism of neuromuscular block [29 (2) vs 53 (9) min, P<0.005] and from antagonism to extubation [6 (1) vs 15 (4) min, P<0.01]. Duration of surgical procedure was the sole multivariate predictor of PORC [odds ratio 0.94 (95% confidence intervals 0.91-0.98), P<0.01]. PORC remains a clinical problem despite use of intermediate-duration neuromuscular blocking drugs and peripheral nerve stimulators. Patients undergoing procedures of short duration may be at risk of inappropriately early tracheal extubation, possibly due to work pressures. The association between suboptimal antagonism of neuromuscular blockade and short procedures needs reinforcement during postgraduate training and departmental quality assurance.