Aims. The paper reports a study to explore the decision process nurses go through before leaving, focusing on leaving decisions that are precipitated by a single, jarring event or shock. Background. Nursing turnover is a significant problem. Although a range of initiatives has been adopted to improve retention, recent insights from the academic literature on labour turnover have additional implications for how this problem might be managed. Method. A structured questionnaire, with some open-ended items, was used to collect data. For respondents who reported a shock ( n = 153), responses were cluster analysed ( hierarchical, agglomerative clustering generated a solution and k-means clustering enhanced the solution). Clusters were validated using responses to open items. Results. There were three broad clusters of nursing turnover: cluster 1 described nurses whose decision to leave was precipitated by a shock that was work-related, negative and unexpected; cluster 2 described those whose decision was precipitated by a shock that was personal, positive and expected; cluster 3 describes those whose decision unfolded more gradually. Cluster 3 described the conventional picture of how turnover occurs (i.e where there is no shock), whereas clusters 1 and 2 were evidence of different types, where a shock prompts the quitting. Conclusion. In many cases of nurse turnover, a single, jarring event, or shock, initiates thoughts of quitting. Understanding the role of shocks has implications for a range of management activities. Allocation of education, promotion and distribution of other benefits should be managed in such a way as to minimize the likelihood of shocks. Profiling of nurse leavers should be undertaken so that managers have an accurate and detailed picture of turnover.