Background Population aging might increase the prevalence of undernutrition in older people, which increases the risk of frailty. Numerous studies have indicated that myokines are released by skeletal myocytes in response to muscular contractions and might be associated with frailty. This study aimed to evaluate whether myokines are biomarkers of frailty in older inpatients with undernutrition. Methods The frailty biomarkers were extracted from the Gene Expression Omnibus and Genecards datasets. Relevant myokines and health-related variables were assessed in 55 inpatients aged >= 65 years from the Peking Union Medical College Hospital prospective longitudinal frailty study. Serum was prepared for enzyme-linked immunosorbent assay using the appropriate kits. Correlations between biomarkers and frailty status were calculated by Spearman's correlation analysis. Multiple linear regression was performed to investigate the association between factors and frailty scores. Results The prevalence of frailty was 13.21%. The bioinformatics analysis indicated that leptin, adenosine 5'-monophosphate-activated protein kinase (AMPK), irisin, decorin, and myostatin were potential biomarkers of frailty. The frailty group had significantly higher concentrations of leptin, AMPK, and MSTN than the robust group (p < 0.05). AMPK was significantly positively correlated with frailty (p < 0.05). The pre-frailty and frailty groups had significantly lower concentrations of irisin than the robust group (p < 0.05), whereas the DCN concentration did not differ among the groups. Multiple linear regression suggested that the 15 factors influencing the coefficients of association, the top 50% were the ADL score, MNA-SF score, serum albumin concentration, urination function, hearing function, leptin concentration, GDS-15 score, and MSTN concentration. Conclusions Proinflammatory myokines, particularly leptin, myostatin, and AMPK, negatively affect muscle mass and strength in older adults. ADL and nutritional status play major roles in the development of frailty. Our results confirm that identification of frailty relies upon clinical variables, myokine concentrations, and functional parameters, which might enable the identification and monitoring of frailty.