In healthcare systems in high-income countries, critical incidents are increasingly seen as an important indicator of the quality of care. Based on the rationale that there are important lessons to be learnt from mistakes and that insights into critical incidents will help to prevent them from happening again, there is a widespread assumption that conducting inquiries will contribute to improvements in patient safety. In this article, we draw on data from a qualitative comparative case study of three critical incidents in Dutch hospitals in the last decade to examine the ways in which critical incidents are investigated. Through a detailed analysis of the inquiry documentation, we identified four key elements in the inquiry process: how risks were framed and perceived, the type of methods the inquiries used to examine critical incidents, the ways in which inquiries allocated blame and the ways in which they sought to maintain transparency. Drawing on Schön and Rein’s work on framing theory, in this article we examined how the key participants in the inquiries framed issues so that they could undertake their work. We found that inquiries are complex processes in which inquiry teams can and do use different frames for deciding who should be involved in the inquiry, what should be discussed, how this should be done and to whom findings of the inquiry should be disclosed. We found that inquiries used professional, managerial or governance frames and sometimes elements of two or more frames coexisted. Within these frames, risk was framed in different ways, leading to different types of actions, involving different groups of actors.