This paper reviews the state of the art of research on individual decision-making in high-stakes, low-probability settings. A central theme of the discussion is that optimally resolving high-stakes decisions poses a formidable challenge not only to naïve decision makers, but also to users of more sophisticated tools such as decision analysis.. Such problems are difficult to resolve because precise information about probabilities is not available, and the dynamics of the decision are complex. When faced with such problems, naïve decision-makers fall prey to a wide range of potentially harmful biases, such as not recognizing a high-stakes problem, ignoring the information about probabilities that does exist, and responding to complexity by accepting the status quo. We offer an agenda for future research focusing on how the process and outcomes of high-stakes decision making might be improved.