OBJECTIVES:  Develop a reliable methodology to correlate records from medical claims databases within the Brazilian public health care system (SUS). METHODS:  Two medical claim databases from the Brazilian Ministry of Health information system (DataSUS) were considered for this study: Ambulatory (SIA) and Hospital (SIH) - both databases are made publicly available in separate without deterministic record keys for record linkage. A record linkage algorithm was developed to craft a broad real world longitudinal patient dataset. A set of parameters such as patient ZIP code, municipality, age or birth date, race, nationality, gender, and ICD were assessed to create a de-identified patient key, as well as to link SIA and SIH datasets. The record linkage methodology consists of a set of eighteen steps based on deterministic and probabilistic connections between de-identified patient keys from both databases; variables are banded into different combinations at each step to maximize the number of connections. Results are considered valid only if no inconsistencies of birth date and gender are found for the same de-identified patient key. Finally, additional variables available were set aside from validation due to reporting inconsistences and volatility. RESULTS:  Linkage outcomes vary depending on disease and health care setting dynamics (e.g. in or out patient) as well as epidemiologic characteristics (e.g. prevalence and age group concentration). As an illustration, within a hepatocellular carcinoma cohort, 1.189 patients were independently found at SIA and 5.140 at SIH as well. Finally, 2.763 patients were linked over the intersection, resulting in a total cohort of 9.092 patients in 2015. CONCLUSIONS:  Both cohort size and the distribution between hospital and ambulatory setting are aligned to published literature providing initial evidences on the potential of the methodology. Such approach promises advances in the development of analysis such as health care resource utilization, hospital admissions, diagnosis and treatment dynamics based on patient-centric real world evidence.