This study examined the effects of social support, poverty, functional status, and demographic variables (age, race, gender) on nursing home admission using Cox proportional hazard models with time varying covariates on the Longitudinal Study on Aging (1984-1990). The Behavioral Model of Health Care Utilization was used to organize the study. Living with a spouse or children reduced the risk as did being socially active. High income reduced the risk but there were problems in the measurement of income. Higher levels of functional disability predicted higher risk. Being Black or being female reduced the risk of admission. Age was a predictor of increased risk. Beyond these findings this paper provides support for the importance of gathering and utilizing data that allows for predictor variables to vary with time.