Objective To explore the potential categories and influencing factors of frailty trajectory in elderly patients on maintenance hemodialysis (MHD) therapy. Methods By using convenience sampling method, elderly MHD patients from three hospitals in Yangzhou were selected. The frailty scores of patients were assessed with the Fried Frailty Phenotype at baseline, 6 months, 12 months and 24 months later. The potential categories of frailty trajectory were identified by the latent class growth model, and the influencing factors were analyzed by logistic regression analysis. Results The trajectory of frailty in elderly MHD patients could be divided into 3 groups:the group with sustained low level of frailty (29.5%), the group with slow increase of frailty (65.5%) and the group with sustained high level of frailty (5.0%). Logistic regression analysis showed that， age, diabetes history, sleep disorder and serum albumin level were influencing factors of the frailty trajectory (all P.<0.05). Conclusion The frailty changes among elderly MHD patients have population based heterogeneity. Intervention measures should be taken in pursuant to the influencing factors.