This prospective study was undertaken to determine the incidence of drug administration errors by anaesthetists at three tertiary South African hospitals. Hospitals A and C treat adults predominantly, whereas Hospital B is a paediatric hospital. Anaesthetists completed an anonymous study form for every anaesthetic performed over a six-month period. They were asked to indicate whether or not an error or near-miss had occurred and if so, the details thereof. A total of 30,412 anaesthetics were administered during the study period. The response rate and combined incidence of errors and near-misses was as follows: Hospital A 48.8% (1:320), B 81.3% (1:252) and C 48.1% (1:250). The overall response rate was 53% and the combined incidence was 1:274. Neither the experience of the anaesthetist nor emergency surgery influenced whether an error occurred or not. Most errors occurred during the maintenance phase of anaesthesia. The most common errors were those of substitution. At the paediatric hospital, incorrect dose was as frequent an error as substitution. Of all errors, 36.9% were due to drug ampoule misidentification; of these the majority (64.4%) were due to similar looking ampoules. Another 21.3% were due to syringe identification errors. No major complication attributable to a drug administration error was reported. Despite an increasing awareness of the problem together with suggestions in the literature to reduce the incidence, drug administration errors remain fairly common in South Africa. Failure to institute suggested solutions will continue to compromise patient safety.