OBJECTIVES:  Hemophilia-related bleeding events are difficult to quantify. Insurance claims’ data may capture the information, but there is no robust methodology to identify these events. METHODS:  Using 2004-2012 Truven Health MarketScan commercial claims database (coverage of >30 million employees in th United States), we evaluated males under age 65 having ≥1 inpatient/≥2 outpatient claims 30-day apart  for hemophilia A or B (ICD-9: 286.0, 286.1), and ≥1 pharmacy claim for clotting factors VIII or IX with ≥12 months continuous enrollment from first hemophilia treatment were selected. Bleeding events were identified using ICD-9 codes for hemarthrosis, hematoma or other acute bleeds in inpatient/outpatient claims. Hemophilia treatment guidelines recommend 1-14 day treatment depending on bleeding severity. We explored aggregating claims within varying time windows (1, 7, 10, and 14 days) to capture a single bleeding event and assessed sensitivity of the time differences using pairwise rank correlations. RESULTS:  We identified 2,425  and 269  hemophilia A and B patients. Using a 7-day window, average annual bleeds ((total bleeds ÷ patient-months)*12) among patients with ≥ 1 bleed equaled 2.31  (range: 1-33) for hemophilia A and 1.92  (range:  1-18) for hemophilia B. Results were similar for a 10-day window (hemophilia A: 2.21 (1-31), hemophilia B: 1.81 (1–14)) and 14-day window (hemophilia A: 2.10 (1–24), hemophilia B: 1.69  (1–12) ) ). A 1-day window produced somewhat higher numbers (hemophilia A: 2.99 (1-66), hemophilia B: 2.72  (1-45)), but pairwise rank correlation remained high across the four assumptions (coefficients ≥ 0.99, p-values <0.01 for both disease types). Frequencies are comparable to a large US-based study reporting bleeds in moderately-severe patients, albeit smaller than studies of severe patients. CONCLUSIONS:  Claims data can be utilized to construct stable, robust indices of bleeding events in hemophilia patients, permitting reliable studies of factors influencing bleeding frequency and healthcare burden.