Clinical studies and the National Kidney Foundation-Dialysis Outcomes Quality Initiative guidelines suggest that a target hematocrit of 33% to less than 36% is appropriate for patient benefit. Previous studies have shown an association of lower risks for death and hospitalization when hematocrits were 33% to less than 36%. In this study, we assessed the relationship between hematocrit value and associated Medicare expenditures, analyzing incident Medicare hemodialysis patients from January 1, 1991, through June 30, 1995. All patients survived at least 90 days to normalize eligibility and an additional 6-month entry period to assess comorbidity and hematocrit values. All patients were followed up from July 1, 1991, through December 31, 1996. We assessed the association between hematocrit values in the 6-month entry period and the Medicare-allowable Part A and Part B per-member-per-month (PMPM) expenditures in the follow-up period, controlling for other variables, including patient demographic characteristics, comorbid conditions, and severity of disease. We found that hematocrits of 33% to less than 36% and 36% and higher were associated with lower Medicare-allowable payments in the follow-up period. Compared with reference patients with hematocrits of 30% to less than 33%, the Medicare-allowable PMPM expenditures were significantly greater for patients with hematocrits less than 27% and 27% to less than 30% (12. 7% and 5.3%, respectively), and the Medicare-allowable PMPMs were significantly less for patients with hematocrits of 33% to less than 36% and 36% and higher (6.0% and 8.2%, respectively). Although these findings suggest that the treatment of anemia may be associated with significant savings in total patient Medicare expenditures, caution should be considered because these findings are associations and should not be deemed as showing causality.