Many studies have documented a longevity advantage for married persons relative to their unmarried counterparts in all age groups. However, these studies have failed to determine whether the advantage experienced by married elderly persons arises mostly from selection and causal processes which operated at younger ages. This paper employs data from the Longitudinal Study of Aging (1984-1990) to explore whether marital status continues to exert any influence on health and mortality at the older ages. In the presence of an extensive set of controls for health status at the baseline survey, a series of logistic models are used to determine: (1) the magnitude of marital status effects on disability and on mortality, among older males and older females; and (2) the extent to which the social environment and economic status of the elderly can account for the existing disability and mortality differences by marital status.