Purpose: Ultrasound-based prediction methods for the detection of low muscle mass for sarcopenia in older adults have been explored previously; however, the muscle that most accurately predicts it is unclear. This study aimed to clarify prediction accuracy and cut-off values for ultrasound-derived thigh and lower leg muscle thickness (MT) to detect low skeletal muscle mass index (SMI) in older adults and to estimate cut-off values based on two standard deviations (SD) below younger adult means for the corresponding muscles as an early detection tool for site-specific low muscle mass. Methods: This study included 204 community-dwelling older (64 males, 140 females, mean age: 75.4 years) and 59 younger (31 males, 28 females, mean age: 22.3 years) adults. The MT of the rectus femoris, vastus intermedius, gastrocnemius, and soleus muscles was measured using ultrasound; SMI was measured using bioelectrical impedance analysis. Results: The prevalence of a low SMI among older adults was 20.3% (n=13) for males and 21.4% (n=30) for females. The receiver operating characteristic analysis revealed that the total MT for the four muscles measured presented the highest area under the curve (AUC) value to predict low SMI for males (0.849) and females (0.776). The AUC value was the highest for the total MT of the gastrocnemius and soleus muscles for males and the gastrocnemius for females (0.836, 0.748; cut-off value: 5.67 cm, 1.42 cm, respectively). Muscle-specific differences between the low SMI-predicting and SD-based cut-off values were observed. The SD-based value for the rectus femoris (1.85 cm) was substantially higher than the low SMI-predicting value (1.51 cm) in males. Conclusion: Ultrasound measurement of lower leg muscles may be a simple, robust measure to detect low muscle mass for sarcopenia. Additionally, cut-off values for site-specific muscle mass loss may not always agree with those for whole-limb muscle mass loss.