There is an increased need for the development of approaches to measure quality, costs and resource utilization patterns among allogeneic hematopoietic cell transplant (allo HCT) patients. Administrative claims data provide comprehensive data for examining service utilization and costs. However, because administrative data are primarily designed for reimbursement purposes, many challenges arise when using it for research. This is particularly problematic when working with rare diseases and procedures, as is the case with allo HCT. To examine the costs and service utilization of allo HCT and chemotherapy for older patients with acute myeloid leukemia (AML), a retrospective cohort study was designed using the 2007-2011 Truven Health MarketScan Research Database, a national multi-payer claims database that contains costs paid for healthcare services. Using this cohort, we demonstrate challenges in using claims databases for allo HCT and propose some approaches to mitigate them. Using ICD-9 diagnosis codes, we identified 29,915 patients with AML in the dataset. From this cohort, we focused on patients age ≥ 50 years and preliminarily identified 468 allo HCT recipients and 6371 patients with AML who did not receive an HCT. Figure 1 lists some of the issues faced and our approach to address them. Utilizing administrative claims data is both a science and an art. Researchers need to make rational decisions based on clinical knowledge but allow adjustment for data limitations to ensure valid results and accurate measures of quality, costs and resource utilization. A thorough understanding of strengths and weaknesses of claims data are necessary in order to minimize bias and to select a cohort that is most representative of the questions that are being addressed. Cohort selection is presently ongoing and additional challenges and solutions will be discussed.