[en] Error detection has not received much attention from the scientists since human error has been shown as the main cause of accident in complex systems. However, reducing the consequences of error depends largely on error detection. The goal of this paper is to synthesize the existing scientific knowledge on error detection, mostly based on studies conducted in laboratory or self reporting and to complete it through the analysis of a corpus of cases collected in a complex system: anaesthesia, in order to better describe how this knowledge can be used to improve our understanding of error detection modes. We used an anaesthesia accident reporting system we developed and organized at two Belgium University Hospitals to collect information about the error detection patterns. Results show that detection of errors principally occurred through standard check. We found significant relationships between the type of error, the type of error detection pattern, and the training level of the anaesthetist who committed the error.