Target controlled infusions (TCI) of propofol allow anaesthetists to target constant blood concentrations and respond promptly to signs of inappropriate anaesthetic depth. Studies comparing propofol TCI with manually controlled infusion (MCI) reported similar control of anaesthesia, but did not use an objective measure of anaesthetic depth. We therefore tested whether the Bispectral Index (BIS), an electroencephalographic (EEG) variable, is more stable during propofol TCI or MCI. Forty patients received midazolam and fentanyl before induction and were randomized to TCI or MCI. Target propofol concentrations in the TCI group were 3 to 8 microg/ml. The MCI group received propofol bolus (approximately 2 mg/kg) and infusion (3 to 10 mg/kg/h). Neuromuscular blockade was achieved with rocuronium. Following endotracheal intubation, nitrous oxide (66%) in oxygen was delivered and propofol infusion and fentanyl boluses were titrated against clinical signs. Blood pressure, heart rate and EEG were recorded, although the anaesthetist was blind to BIS values. The ideal BIS for general anaesthesia was defined as 50. Performance error, absolute performance error, wobble and divergence of BIS, and maximum changes in blood pressure and heart rate were compared using two-sample t-tests or rank-sum tests where appropriate. There was no difference in absolute performance errors during maintenance of anaesthesia with propofol TCI or MCI (23 +/- 11% vs 23 +/- 9%; P=0.97). The two groups did not differ significantly in performance error, wobble, divergence on haemodynamic changes. We conclude that TCI and MCI result in similar depth of anaesthesia and haemodynamic stability when titrated against traditional clinical signs.