<i>Background:</i> Increased mortality risk at low body mass index values is well established for older persons. It is, however, unclear how the underlying body mass components (fat and muscle mass – FM and MM, respectively) are associated with mortality in old age. <i>Objective:</i> This study aimed to examine the mortality risk of four body composition measures (appendicular skeletal MM, leg, arm and trunk FM) with 12-year mortality in community-dwelling older men and women. As a secondary objective, the influence of cancer, obstructive lung disease, smoking and previous weight loss on these associations was examined. <i>Methods:</i> Data were used from the Longitudinal Aging Study Amsterdam, a random population-based cohort study (55–85 years) in the Netherlands. Body composition was determined in 1995–1996 by dual energy X-ray absorptiometry. The present study included 477 community-dwelling persons aged ≧65 years who were followed until 2007 for their vital status. <i>Results:</i> Twelve-year mortality rates were 133/242 (55%) in men and 92/235 (39%) in women. Since most associations were U- or J-shaped, only observations below the sample mean were included to calculate hazard ratios (HRs) per one SD lower value. Adjusted for height, age and each other, lower appendicular skeletal MM [HR 1.59 (95% CI: 1.04–2.42)] and lower leg FM [1.68 (1.04–2.72)] in men and lower trunk FM [1.61 (1.02–2.53)] in women were associated with an increased mortality risk. Associations attenuated and became statistically nonsignificant in men after adjustment for cancer, obstructive pulmonary disease and smoking and in women after additional adjustment for previous 3-year weight change. <i>Conclusions:</i> In older men, lower MM and lower leg FM are associated with an increased mortality risk, while in older women only lower trunk FM is associated with an increased risk. The causality of these associations is debated. Suggested sex differences deserve further study.