This article presents a research project which aims to examine error management in emergency medical assistance, placing it within the broader framework of risk management. The data mainly come from the activity of operators at an emergency call desk who deal with requests for emergency medical care and decide what type of assistance to send to victims. The results show that operators make errors, particularly violations. The operators tolerate these violations and many errors are left undetected. Most of the errors that are detected are not rectified by the operators themselves, but mostly by third parties. Any protection put in place against errors, particularly against any risks, was observed. Despite the errors, the system was considered safe, at least during the observed period. Error and violation management has its place in risk management activity. The discussion attempts to model this activity, which is based on trade-offs between security objectives (managing external and internal risks that could lead to the death of casualties due to lack of adequate assistance) and economic objectives (limitation of system costs associated with the implementation of unnecessary or excessive assistance). Arbitration tends to be for the benefit of safety and the objective is not to optimize assistance sent to individual victims but rather to provide assistance to all victims. Arbitration in favour of security is revealed through three main activity characteristics: 1) in the medium-term, protection against errors and especially against risks, aimed at avoiding the most dreaded events and to anticipate unwanted events depending on the evolution of the situation (first safety net); 2) in the short-term, production (and tolerance) of violations that occur as risks taken (accepted) to improve target performance; and 3) recovery of errors or violations (or consequences) not managed by the operators themselves, but instead left to others (second safety net). The discussion also opens perspectives on approaches to security which focus on the situation under study.