The study of unsolicited psychiatric patients who became ill because of their experience in a natural disaster can assist in the design of future disaster research. A clinical report of 36 such patients illustrates the problems of case detection, the delayed presentation of much of the morbidity, and the need to separate stress-related symptoms which are common in disaster victims from psychiatric illness. Unless these issues are taken into account, estimates of the prevalence of psychiatric disorders after major disasters may be subject to substantial error. The role of vulnerability factors assessed to be operating in these patients suggests that exposure and losses sustained in the disaster alone are inadequate predictors of psychiatric disorder. The risk factors for the development of disaster-related psychiatric morbidity will be more accurately defined if the contribution of a range of constitutional, personality, and social factors as well as the personal impact of the disaster are investigated in future research.