OBJECTIVES: To evaluate the Medicaid Analytic Extract (MAX) database for use in pharmacoepidemiologic research using hemophilia as a test case. METHODS: This research was conducted using MAX data (years 2000 - 2004) for 49 states (excluding AZ) and DC. We identified patients who received recombinant factor VIIa (rFVIIa), factor VIII inhibitor bypass activity (FEIBA), factor VIII or factor IX or had a diagnosis code for coagulation defects (ICD-9 286.xx). Use of antihemophilic drugs was identified through either pharmacy dispensing or provider administration codes. Several definitions were applied to identify patients with hemophilia A and B. Health care utilization and costs for each year were characterized for rFVIIa/FEIBA users. RESULTS: After excluding 5 states with extensive use of managed care (for which claims may be incomplete), a source population of approximately 63 million people was identified.  A total of 130,946 patients met the inclusion criteria, of which 10,693 had at least one diagnosis for hemophilia A and 2,032 for hemophilia B. Fewer than 10% were enrolled in managed care in any year. Restricting to males reduced the sample size to 7292 for hemophilia A (mean age 15.4 years) and 1557 for hemophilia B (mean age 13.8 years), and these numbers correspond to the expected relative frequencies of these disorders (4:1). Users of rFVIIa or FEIBA were approximately 5% of the cohort and these patients were characterized by extensive use of health care with on average, a physician visit every 6 weeks and yearly total costs approaching $1,000,000. CONCLUSIONS: As with any administrative data source, researchers should be aware of limitations within MAX data, such as the validity of diagnostic codes and the potential incompleteness of data for certain population subgroups. Broad demographic coverage, however, large size, and extensive representation of children make MAX useful, particularly for studying low-prevalence conditions such as hemophilia.