The general aim of this research is to use the multistate approach to active life expectancy to provide a fuller portrayal of the interaction of health and mortality among the older members of the American population. The impetus for this study comes from a desire to clarify how differences in mortality and morbidity conditions affect the health status of a population and the expected life cycle experiences of individuals. Given the current stratification of morbidity and mortality rates across sociodemographic groups, our analysis attends to major cleavages in active life expectancy among the oldest members of the American population. Our strategy to examine these issues rests on the following specific aims: 1) the estimation of levels of active and inactive life expectancy during the latter half of the 1980's. In this phase of the study, we question the utility of previously employed definitions of active and inactive life and propose a division of life expectancy into states based on ability to provide a given level of self-care. These definitions divide life when functioning is less than complete into a set of policy-relevant health states. 2) the use of multivariate hazard models to probe the social, demographic and medical correlates of changes in health status in old age; and 3) to derive active life tables from the multivariate hazard models to achieve a substantive interpretation of the implications of the combined set of hazard-model estimates for older persons' health expectancy and for population health. This will result In life tables for more detailed subgroups of the population than has previously been possible. Further, this approach will allow the identification of how sociodemographic characteristics shape the potentially complex processes by which older individuals experience both improvement and declines in health. The Longitudinal Study of Aging is the major data set to be employed in this research.