The objective of the four studies presented here was to understand how risk is managed during the whole anesthesia process, using different methods. Risk prevention and supervision of normal and abnormal situations were studied. The first exploratory study aimed to understand incidents in anesthesia using interviews based on the Critical Incident Technique (Flanagan, 1954). Eight experienced anesthetists had to describe their activity, and to relate one or more negative and positive incidents. Two kinds of significant factors determine the occurrence of an incident: human and organizational. An incident appears to be the consequence of a sequence of events, showing the causal complexity of an incident. Finally, the anesthetists? work is focused on risk management with the forecasting of future problems. The second and third studies were devoted to an examination of risk assessment and prevention. In the second study, 25?preoperative reports were analyzed and five experienced anesthetists participated. The information contained within the preoperative reports relates to risk factors assessment and risk prevention. The second category includes prevention strategies; these are mostly warnings used to manage actions during surgical intervention. Finally, the preoperative consultation is mainly used to estimate and anticipate risks. In the third study, 20?anesthetists (10?experts and 10?novices) had to categorize?16 patient files that were constructed according to the difficulty level of the patient?s health state and the difficulty level for surgical intervention. Experts form more categories than do novices. Anesthetists, irrespective of the expertise level, form categories using three kinds of criteria: surgery, patient and anesthetic risk level. Experts mainly use the patient criteria to categorize patient files, whereas novices mainly use the anesthetic risk level. Finally, the fourth study was devoted to the analysis of anesthetists? supervision in the operating room. Twenty-two surgical interventions were observed. During surgery, a large number of drifts occurred (72?% of incidents), all of which were recovered. To categorize incidents, the ants system was used (Flin, Fletcher, Galvin, Maran, & Patey, 2003). Most of the incidents were due to inappropriate situation awareness. Anticipation failures and defaults in information recognition are the main problems. Three main kinds of actions were observed: task management improvement by providing and maintaining standards, better decision making in balancing risks and selecting options, and situation awareness improvement from gathering information differently. The main discussion points out the importance of studying the recoveries of small deviations. In a dynamic process, the preparation is mostly devoted to risk evaluation; risk prevention is mainly carried out by bringing to light any potential problems. Lastly, the importance of real work observation in the study of resilience is noted.