We tested the hypothesis that a high concentration of serum ferritin, a frequently used marker of iron stores in dialysis patients and an acute-phase reactant, may be a marker of morbidity and mortality in these patients. To evaluate the impact of ferritin on morbidity and mortality, we reviewed the 6-month hospitalization rates in our dialysis patients retrospectively and subsequently reviewed the mortality among these patients over a 12-month period of time prospectively. One hundred one adult hemodialysis patients (59 men and 42 women; age, 54 +/- 15 years) who had been on hemodialysis for 38 +/- 27 months were studied. All but 5 patients were on intravenous iron with similar iron administration pattern. In the retrospective cohort, ferritin's correlation coefficients for hospitalization days and frequency (both r = +0.39, P: < 0.001) were higher compared with the albumin correlations for hospitalization days (r = -0.31, P: = 0.001) and frequency (r = -0.28, P: = 0.005) and correlation coefficients remained similarly significant after case-mix adjustment. In the prospective study, the "predeath" value of serum ferritin for 17 deceased patients (891 +/- 476 ng/mL) was higher than both their "initial" value (619 +/- 345 ng/mL, P: = 0.007) and the mean ferritin value of 84 surviving and withdrawing patients (639 +/- 358 ng/mL, P: = 0.001). Although Cox proportional hazard analysis showed a significant odds ratio of death only for serum albumin and not for ferritin, logistic regression analysis using the predeath values confirmed the significant impact of both decreased serum albumin and increased serum ferritin as markers of dialysis mortality. After case-mix adjustment, the relative risks of death for a 500 ng/dL increase in serum ferritin was 2.71 (95% confidence interval, 1.06 to 7.02) and for a 0.5 g/dL decrease in serum albumin was 4.48 (95% confidence interval, 1.77 to 11.33). Hence, serum ferritin is a strong predictor of hospitalization in dialysis patients. Although serum albumin is found to be a strong long-term marker of mortality in hemodialysis patients, an increase in serum ferritin appears to be a more reliable short-term marker of death over a 12-month period. Therefore, in the setting of uniform iron administration, a high serum ferritin may be a morbidity risk factor and a recent increase in serum ferritin may carry an increase in the risk of death in these patients.