An estimate of total body muscle mass with dual-energy X-ray absorptiometry (DXA; appendicular muscle mass divided by 0.75) was compared with 24-h urinary creatinine excretion in 59 healthy men and women [20-30 yr (younger), 45-59 yr (middle age), and 60-79 yr (older)] who stayed in a clinical research center for 5 days. Total body water ((2)H(2)O dilution), fat (underwater weighing), bone mineral (DXA), and total body protein mass (based on a 4-compartment model) were also measured. Muscle mass estimates by DXA and creatinine were highly correlated (r = 0.80). However, stepwise multiple regression indicated that a significant amount of additional between-subject variability in DXA-based muscle mass estimates could be explained by total body water. Creatinine excretion, knee extensor strength, and total body protein mass all decreased with age, suggesting a decline in muscle cell mass with aging. However, DXA-based muscle mass and measures of nonfat body mass (i.e., lean body mass by (2)H(2)O and fat-free body mass by underwater weighing) did not change with age. These results indicate that DXA and urinary creatinine excretion give different results regarding the decline in total body muscle mass with aging. The factor(s) responsible for the apparent underestimate of age-related sarcopenia by DXA remain to be fully defined, but changes in body water may be an important contributor.