Recently reported analysis of the 1999 National Long Term Care Survey confirms the trend that disability levels continue to fall at an increasing rate among the elderly in the U.S., with the disability level falling more rapidly for blacks than for non-blacks between 1994 and 1999. Yet despite these gains blacks still suffer higher levels of disability compared to non-blacks. Geriatric rehabilitation is one of the health care interventions that has been credited with slowing functional decline in the elderly, and concomitantly lowering health care costs. In 1993 the University of Pennsylvania School of Nursing opened the CARE Program, a comprehensive, short-term outpatient rehabilitation program that targeted elders at risk for nursing home or hospital admission due to functional decline and intervened to strengthen their level of functioning in order to stave off institutional care. Just over two-thirds of the patients enrolled in CARE were minorities, and three-quarters were the product of community-based referrals. This pilot study will assess the role of race and other subject characteristics in the outcomes of patients enrolled in CARE when compared to matched control groups drawn from the 1994 National Long Term Care Survey. The specific aims are: (1) To derive control groups from the 1994 NLTCS that are matched to CARE patients and assess the efficiency of the matches. Statistically-matched control groups will be drawn from the NLTCS using propensity scores and random-order nearest-pair methods. (2) To test for differences in hospital use between the CARE patients and the NLTCS matched control groups, and assess the contributions of race and other characteristics in differential outcomes. Proportional hazards regression models will be used to assess differences in the time to hospital admission between CARE patients and controls over a one-year period. Reducing the level of chronic disability has significant and broad implications for the quality of life for the nation’s elders as well as for the solvency of Medicare and Social Security. Racial disparities in disability levels are not well understood, and this study provides an opportunity to explore this issue systematically through the use public-use data sets that can provide a control group where one is lacking.