We describe a research project in which we evaluated the concordance of two different sources for hospitalization data and identified specific factors that were associated with concordance. The first data source was from a longitudinal cohort study that collected information on hospitalizations using patient self-report and active hospital surveillance. The second data source was administrative claims from the Medicare program in the United States. Each data source has unique strengths and weaknesses, and our results have implications for how these data sources can best be used. We merged these data sources and used regression analysis to evaluate whether particular factors were associated with concordance. We encountered challenges including differences in how hospitalizations were categorized in each data source and inaccuracies in discharge dates used to match records. Lessons learned from this study are that it is important to have a deep understanding of how each source collects utilization data, to be aware of how the policy context may affect reporting of utilization, and to verify the accuracy of matching criteria. In conclusion, we recommend this approach for similar research projects to evaluate concordance of two sources for utilization data when no “gold standard” data are available, most events will be captured by at least one (but likely not both) of the data sources, and factors that are potentially related to concordance are also available in the databases.