The aim of this study was to examine the performance of anaesthetists while managing simulated anaesthetic crises and to see whether their performance was improved by reviewing their own performances recorded on videotape. Thirty-two subjects from four hospitals were allocated randomly to one of two groups, with each subject completing five simulations in a single session. Individuals in the first group completed five simulations with only a short discussion between each simulation. Those in the second group were allowed to review their own performance on videotape between each of the simulations. Performance was measured by both 'time to solve the problem' and mental workload, using anaesthetic chart error as a secondary task. Those trainees exposed to videotape feedback had a shorter median 'time to solve' and a smaller decrease in chart error when compared to those not exposed to video feedback. However, the differences were not statistically significant, confirming the difficulties encountered by other groups in designing valid tests of the performance of anaesthetists.