In 1997, the Health Care Financing Administration Hematocrit Measurement Audit (HMA) program initiated use of a 3-month rolling average hematocrit (Hct) level for reimbursement of epoetin claims in hemodialysis patients, with denial of payment when this value exceeded 36.5%. This study evaluated the impact of the HMA program on anemia-related outcomes in hemodialysis patients. An observational, retrospective study of 987 hemodialysis patients from 11 dialysis centers in the United States was performed, collecting data between October 1996 and December 1997. Centers were selected from a pool of nearly all facilities in the United States, which during May 1997 satisfied one of two criteria: greater than 75% of patients at the facility had mean Hct level of > or =33% (Group A) or fewer than 50% of patients at the facility had mean Hct level of > or =33% (Group B). Each facility maintained its own anemia management practices without specific anemia management interventions as part of this study. Hct level, hemoglobin (Hb) level, and epoetin dose were analyzed to compare the pre-HMA period (October 1996 to May 1997) to the HMA period (June to December 1997) and/or for each of the five quarters of the study period. The primary study endpoint was the percentage of patients with Hct levels of > or =33% during each study quarter. The mean Hct level at baseline was 34% in Group A and 33.4% in Group B (p = 0.01). Hct levels, which were increasing before implementation of the HMA program, decreased during the HMA period (p < 0.001 and p = 0.013 in Groups A and B, respectively). The percentage of patients in Groups A and B with mean quarterly Hct levels of > or =33% decreased during the last quarter of the HMA implementation period compared to the quarter immediately preceding the start of the HMA program (p < 0.001 for both comparisons). Changes in Hb levels were similar to those seen in Hct levels. The mean epoetin dose administered decreased from 13,090 U/week at the start of the study to 11,884 U/week immediately before the HMA program took effect (p < 0.05). The HMA program adversely affected anemia treatment outcomes, regardless of whether dialysis units before HMA implementation had <50% of patients with a Hct level of > or =33% or had >75% of patients with a Hct level of > or =33%. The decline in mean weekly dose of epoetin was likely a result of withholding doses out of concern among providers about risk of reimbursement denial.