This study was undertaken to identify variables that could explain the association between low albumin and a 9- to 12-year mortality follow-up among 287 community- dwelling and 176 institutionalized people aged 60 years and over. A wide array of nutrition assessment variables was simultaneously examined in this population to identify confounders of the association. The results show that the risk of mortality for subjects with albumin values of 40 g/liter and over was 0.46 of the risk for those with albumin values below 40 g/liter, after controlling for the confounders age, blood urea nitrogen, triglyceride, history of diseases, and inability to shop owing to medical conditions. Similarly, albumin was also inversely associated with mortality among institutionalized subjects even after controlling for the confounders age, sex, blood urea nitrogen, transferrin, and history of stroke. However, the association was no longer significant among the institutionalized population once the deaths that occurred within the first 3 years after study participation were eliminated. The results indicate that albumin is a long-term predictor of mortality among noninstitutionalized subjects and increased mortality is not only a result of age, history of chronic diseases, medication use, or protein intake. Among institutionalized subjects, albumin appeared to be a short-term predictor of mortality.