OBJECTIVE: To determine if nutritional parameters and discharge setting are associated with mortality in older male veterans on a Geriatric rehabilitation unit (GRU). DESIGN: Two-year follow-up of sequential admissions to the GRU who had laboratory studies completed on admission to the acute hospital and on transfer to and discharge from the GRU. SETTING: University-affiliate VA medical center. PATIENTS: Eighty-three consecutive older male veterans admitted to the GRU. MEASUREMENT: Mortality during a period of 2 years post-discharge. MAIN RESULTS: In a univariate analysis, predictors of mortality were serum albumin level less than or equal to 3.5g/dL on admission to the GRU (P = .01), moderate or severe Nutritional Status Score (P = .03), discharge to a place other than home (P = .01), and use of antibiotics while on the GRU (P = .05). Discharge albumin remained the single significant predictor of mortality in a multivariate analysis (P = .005). CONCLUSIONS: Serum albumin is the strongest predictor of 2-year post-hospitalization mortality of older patients cared for on a GRU. Other predictors were presence of infection, Nutritional Status Score of moderate or severe compromise, and discharge to a place other than home.