The study analyzed the dynamic trajectories of hemoglobin, log-transformed ferritin and phosphorus in 254 maintenance hemodialysis (MHD) patients with 814 repeated measurements. Their associations with all-cause mortality were assessed using joint linear mixed effects modeling and Cox models. During a median follow-up of 18 months, 62 patients (24.4%) died. Higher hemoglobin levels were independently associated with lower mortality (10 g/L increase: HR = 0.97; 95% CI: 0.95–1.00). Log-ferritin had a negative but insignificant association, phosphorus did not improve the model. The best resolution was achieved by the hemoglobin + log-ferritin model (AUC ≈ 0.70–0.75). Sensitivity analyzes with phosphorus or transferrin saturation confirmed the robustness of the results. Dynamic hemoglobin changes provide a stable predictive value for risk stratification and individualized anemia treatment.