We aimed to establish the frequency and nature of drug administration error in anaesthesia (a significant subset of error in medicine) at two hospitals. Anaesthetists were asked to return a study form anonymouslyfor every anaesthetic, indicating whether or not a drug administration error or pre-error (defined as any incident with potential to become an error) had occurred. Further details were sought if the response was affirmative. From 10,806 anaesthetics, 7794 study forms were returned, representing response rates of 80% from Hospital A and 57% from Hospital B (72% overall). The frequency (95% confidence intervals) of drug administration error; of any type, per anaesthetic was 0.0075 (0.006 to 0.009), of i.v. bolus errors was 0.005 (0.0035 to 0.006) and of pre-errors was 0.004 (0.003 to 0.005), with no significant difference between hospitals. Overall, one drug administration error was reported for every 133 anaesthetics. The two largest individual categories of error involved incorrect doses (20%) and substitutions (20%) with i.v. boluses of drug. Of the i.v. bolus substitutions, 69% occurred between different pharmacological classes. One patient was aware while under muscle relaxation, and two required prolonged ventilation. In addition, 47 transient physiological effects were reported, of which five required intervention. We conclude that drug administration error during anaesthesia is considerably more frequent than previously reported.