To determine the incidence of residual neuromuscular blockade after cardiac surgery in patients receiving either rocuronium or pancuronium for muscle relaxation.In a prospective, controlled, double-blind study, 20 patients undergoing coronary artery bypass were randomized to receive either rocuronium (n = 10) or pancuronium (n = 10) during surgery. Anaesthesia was induced with sufentanil, benzodiazepine and propofol or ketamine, and maintained with air/O2/sufentanil/isoflurane. Neuromuscular blockade was induced with 0.1 ml.kg-1 from blinded syringes containing recuronium (6 mg.ml-1) (Group R) or pancuronium (1 mg.ml-1) (Group P). Relaxants were administered according to clinical criteria and reversal agents were not given. After surgery, neuromuscular transmission was assessed by train-of-four stimulation of the ulnar nerve/adductor pollicis EMG (Datex Relaxograph). Mean values from three trains of stimuli were recorded and repeated 30 min later if TOF ratio was < 0.7. Time to extubation was recorded.On arrival in the ICU, nine of 10 patients in group R but only three of 10 patients in group P demonstrated four visible responses (P < 0.05). Mean TOF ratio in group P, 0.03 +/- 0.05, was less than in group R, 0.68 +/- 0.34 (P < 0.001). All patients in group P and 4 of 10 patients in group R had TOF ratio < 0.7 (P = 0.01). Time to extubation in group P (median 18, range 6-48 hr) was not statistically different from that in group R (14, 5-44 hr).Residual neuromuscular block, TOF ratio < 0.7, is common after cardiac surgery but the incidence is less when pancuronium is replaced by rocuronium.