OBJECTIVES: Both inpatient data and outpatient data are logically required to study the continuum of care before, during, and after hospitalization. To create such a repository of longitudinal patient-level information, data extracted from a large US hospital database were matched to a large single-payer claims database. The purpose of this methodological study was to determine which steps in the method were influential toward optimizing the quality and quantity of matched patient data. METHODS: Hospital data were drawn from the Premier Perspective™ Database, a large de-identified US hospital clinical and economic database developed for quality and utilization benchmarking. Claims data were drawn from a large national health plan and provided by i3 Pharma Informatics. De-identified inpatient discharges and outpatient services rendered from January 2005 through December 2007. Inpatient hospitals were matched to the health plan hospitals on Medicare provider number and provider name and address. All claims (hospital, physician, lab, pharmacy) during the timeframe were retrieved and matching episodes of care were defined as claims-level “discharges” based on the available list of variables: provider number, admit date, discharge date, DRG or MS-DRG, gender, and birth date. RESULTS: The inpatient database contained 447 of the 730 matched providers, representing 16.7 million discharges (not limited to the claims data payer). Initial matching linked 355,590 claims-level discharges to 389,482 hospital-level discharges based on the definition of “discharge”. Removing duplicate matches yielded 335,481 one-to-one matched discharges, representing 275,138 unique patients. The final linked and cleaned data extract contains 331,814 discharges, representing 275,138 unique patients. CONCLUSIONS: A series of detailed steps and interim quality checks optimizes the quality and quantity of matched patients for use in subsequent outcomes research. Linking the hospital data to claims data leverages both datasets to enable more robust understanding of patient-level continuum of care than non-linked data.