OBJECTIVES: Real-world evidence informs product development, pricing, and reimbursement as well as patient care. An ongoing challenge, however, is in identifying and accessing the best available evidence to address a given research question. Our objective was to develop a systematic methodology to identify observational data sources for specific needs and to test it in both common and rare conditions in a range of therapeutic areas. METHODS: Systematic literature and Web searches, supplemented with email and telephone contact with data owners, were used to identify and characterize data sources suitable for use in 6 observational research programs spanning oncology, cardiology, and respiratory medicine. Data were captured and project-specific evaluation criteria applied regarding: study population; target geographies; type of clinical, diagnostic, and healthcare data required; length of follow up; and ability to identify patients across data sources. Data sources best meeting program objectives were recommended.  RESULTS: Over 14,000 references were screened across 6 separate studies. Of these, 4.5% to 8.6% were relevant to each study and were reviewed further. Direct contact with data owners was often needed to clarify whether inpatient and outpatient drug utilization, laboratory and imaging results, clinical assessments, healthcare utilization, patient-reported outcomes, and mortality data were available. Detailed characterization of the data sources identified by the systematic search reduced their number by 150–300-fold. This generated a manageable number of data sources (4–17 per study) to address specific research questions, including for prevalent cardiac conditions as well as a rare respiratory sub-population, in target geographies. CONCLUSIONS: The approach was effective in identifying accessible, relevant, and high-quality data for both rare and more prevalent conditions. A systematic understanding of real-world evidence has helped to guide observational research programs in diverse therapeutic areas with specialized data requirements.