The present dissertation is a report of a search process to find out if the level of variance explained typically reported in longevity studies can be increased. Two premises form the rationale of this endeavor: 1. the methods used in most studies of longevity may be improved-upon; 2. the data used to predict longevity in most studies may not be the most appropriate with regard to the target population: those persons indicated as "elderly". In the course of this search process, several new concepts and empirical approaches will prove to be useful. If the search process has a positive outcome, it will offer guidelines to investigators planning longevity studies in the near future. If its outcome turns out to be negative, the lack of predictive ability of longevity studies may be a basic fact which can no longer be ignored and which has consequences for our thinking about and interpretation of predictors of longevity. This dissertation, in other words, deals with the feasibility of predicting longevity in the elderly, from a concern with both the scientific value and the practical applicability of findings. In terms of the typology described above, the research constituting the search process follows the approach of the second category: basically descriptive studies that attempt to identify groups at risk.