BACKGROUND There is an inverse relationship between the soluble interleukin-2 receptor (sIL-2R) and serum albumin, cholesterol, transferrin, prealbumin, and hemoglobin. Inasmuch as low serum albumin and cholesterol have been associated with excess mortality, we hypothesized that elevated sIL-2R would predict mortality in older adults. OBJECTIVE To determine if elevated sIL-2R predicts mortality in patients on a geriatric rehabilitation unit. METHODS Prospective cohort. METHODS University-affiliated VA medical center. METHODS Seventy-two male patients aged greater than 60 years admitted to a geriatric rehabilitation unit. Patients with severe hepatic or renal disease were excluded. METHODS We measured serum albumin, prealbumin, cholesterol, transferrin, hemoglobin, body mass index (BMI), C-reactive protein (CRP), and sIL-2R upon admission. Subjects were followed for 1 year. RESULTS Low serum albumin, prealbumin, and hemoglobin and high sIL-2R and CRP predicted 1-year mortality on univariate analysis. When these predictors were included as covariates in a Cox regression model, only sIL-2R was a significant independent predictor of mortality (P = .043). Multiple linear regression with the above covariates revealed that only sIL-2R predicted time to death at (P = .003). CONCLUSIONS High sIL-2R and CRP and low albumin, prealbumin, and hemoglobin predicted mortality using univariate analysis on a rehabilitation unit. However, with multivariate analysis, sIL-2R was the sole predictor of mortality.