Sarcopenia, characterized by low muscle mass and function, results in frailty, comorbidities and mortality. However, its prevalence varies according to the different criteria used in its diagnosis. This cross-sectional study investigated the difference in the number of sarcopenia cases recorded by two different measurement methods of low muscle mass to determine which measurement was better. We recruited 878 (54.2% female) individuals aged over 65 years and obtained their body composition and functional parameters. Low muscle mass was defined as two standard deviations below either the mean height-adjusted (hSMI) or weight-adjusted (wSMI) muscle mass of a young reference group. The prevalence of sarcopenia was 6.7% vs. 0.4% (male/female) by hSMI, and 4.0% vs. 10.7% (male/female) by wSMI. The κ coefficients for these two criteria were 0.39 vs. 0.03 (male/female), and 0.17 in all subjects. Serum myostatin levels correlated positively with gait speed (r = 0.142, p = 0.007) after adjustment for gender. hSMI correlated with grip strength, cardiopulmonary endurance, leg endurance, gait speed, and flexibility. wSMI correlated with grip strength, leg endurance, gait speed, and flexibility. Since hSMI correlated more closely with grip strength and more muscular functions, we recommend hSMI in the diagnosis of low muscle mass.