This study proposes a primary screening system for diagnosing sarcopenia in older adults through anthropometric measures. This exploratory research initially involved 150 elderly individuals, of whom 122 were selected after a data purification process. Using machine learning techniques such as hierarchical clustering and decision trees, the original set of 13 anthropometric measures was reduced to five key features. Three classification systems were created: the first based on previously established parameters (appendicular muscle mass, walking speed, and grip strength); the second considered upper limb measures (average muscle mass of both arms, grip strength, walking speed, and body fat percentage); and the third focused on lower limb measures (average muscle mass of both legs, grip strength, walking speed, and body fat percentage). These classification systems were clinically validated in a group of 57 patients previously diagnosed by specialists, of which 10 received a positive sarcopenia diagnosis. The results showed similar efficiencies in all three systems, with eight of the ten known positive diagnoses classified in the same group. Additionally, the study provides specific cut-off points for each system, thus facilitating the clinical diagnosis of sarcopenia by medical professionals.