BACKGROUND AND AIMS: The aim of this study was to verify the hypothesis that a combined measurement of albumin and HDL-C might predict total mortality in institutionalized frail older residents. METHODS: Participants were 344 older subjects (272 F, 72 M), living in the "Istituto Riposo Anziani" (I.R.A.), a nursing-home located in Padova, North-east Italy. Functional status, comorbidity, and clinical chemistry parameters were evaluated at entry. All-cause mortality was evaluated after 2 and 4 years. The sample was divided into 4 groups by using the 50 degrees percentile of albumin and HDL-C as cut-off value. The mortality odds ratio (OR) was estimated by multivariate logistic regression analysis. RESULTS: Total mortality was 36.8% after 2 years and 51.8% after four years. A trend toward an increase in mortality from group 1 to 4 was observed (p for trend: 0.01). The OR for 2 and 4 years mortality was 3.83 (95% CI 1.86-7.58) and 2.66 (95% CI 1.37-5.17), respectively, in group 4 compared with group 1, after adjustment for age, gender, number of chronic diseases, functional status, BMI, diabetes, dementia, stroke, CHD, CHF, hypertension, depression, COPD, and total cholesterol levels. CONCLUSIONS: Among frail older nursing-home residents, simple measurement of serum albumin and HDL-C levels may be useful in identifying varying degrees of frailty.