OBJECTIVE To examine survival and community discharge outcomes related to rehabilitation services among patients admitted to nursing homes before the implementation of the Balanced Budget Amendment of 1997. METHODS Retrospective cohort. METHODS A total of 945 Medicaid-certified nursing homes in Ohio. METHODS A total of 11,150 patients admitted for the first time to a nursing home from 1994 to 1996. METHODS Not applicable. METHODS Community discharge and survival rates among patients who did or did not receive rehabilitation services, using multivariable techniques to adjust for patients' propensity to receive rehabilitation and for other potential confounders. In secondary analyses, we also examined dose-response effects and analyzed the effects of rehabilitation when patients were divided into 5 diagnostic groups (stroke, hip fracture, congestive heart failure, chronic lung disease, other). RESULTS Rehabilitation was provided to 58% of the patients and was associated with higher community discharge rates (relative risk=1.48; 95% confidence interval [CI], 1.40-1.57) and a lower hazard of death (hazard ratio=.81; 95% CI,.75-.88). Dose-response effects were observed for both outcomes (P<.001) among patients receiving rehabilitation. Rehabilitation was associated with improved community discharge rates in each of the 5 diagnostic groups. CONCLUSIONS New reimbursement policies that discourage the provision of rehabilitation services may have adverse effects on patients, their families, and societal costs of care.