How can we analyse real decision-making (DM) processes in organizations to prevent or mitigate industrial risks? The main purpose of the proposal is to acquire a better understanding of DM’s psycho-social mechanisms in organizations to improve them and so that, mitigate and/or prevent risks.Managing risk is a priority for ensuring safety and continuous performance. Reason (Reason, 2000) shows how accidents happen when a series of ‘threats’ to safety occur and pass unnoticed through ‘gaps’ in an organization protective system. The actual sequence of actions and decisions in hazardous operations can help to understand, and therefore prevent or mitigate such propagation of failures. However, this needs to document it, and gathering such data is a gap in many organizations. Our research and operational question is therefore: “How can we gather and study real-world data on decision-making in risky situations?”Following the flight recorders’ example, this research project proposes to develop a method to collect DM in hazardous situations, based on a human-centered “cognitive black box”. It will develop an in situ collect and analysis kit for experts’ DM and thinking process by combining a wearable first-person perspective video capture device, called subcam, and protocol for on-the-fly capture of workers’ mental processes. This technique will be used to investigate some real-world cases (in LSE ethics committees and EDF managers’ meetings). The main decision-makers will be equipped with the subcam, then debriefed.The proposed research is a breakthrough in psychology and decision science. Our approach is radically innovative in terms of risks prevention and mitigation in that it not only provides a new path for scientific investigation, but opens the way to industrial applications in many industries, through an innovative and transdisciplinary approach in social sciences.