The performance of a modified target-controlled infusion system was investigated in 16 dogs undergoing routine dental work, by comparing the predicted concentrations of propofol in venous blood samples with direct measurements; the optimum targets for the induction and maintenance of anaesthesia were also identified. The performance of a target-controlled infusion system is considered clinically acceptable when the median prediction error, a measure of bias, is not greater than ±10 to 20 per cent, and the median absolute performance error, a measure of the accuracy, is not greater than 20 to 30 per cent. The results fell within these limits indicating that the system performed adequately. The optimal induction target was 3 μg/ml, and anaesthesia of adequate depth and satisfactory quality was achieved with maintenance targets of between 2.5 and 4.7 μg/ml propofol. The system was easy to use and the quality of anaesthesia was adequate for dental work.