OBJECTIVES: Longitudinal Patients Database (LPD) is a primary care database of anonymized electronic medical records (EMR) from about 4 % of the French population. Diagnosis and prescription data are routinely collected from proprietary practice management software used by physicians (primary care and specialists) to maintain EMR of their patients. Although LPD has been extensively validated by numerous publications and its use by French National Health Authorities, this is the first time that its representativeness and validity is systematically examined. METHODS: The distribution of several variables were analyzed and compared to available literature. Part of these variables refers to physician’s practices participating to the database while others refer to patients in these practices. Data about prevalence, treatments, and patients profile were retrieved from published French Health Authorities studies based on LPD data and compared to other published sources. RESULTS: The sampling methods for the physician’s selection practices were shown to provide a good representativeness of the physician panel. Analyze of the patients population showed that LPD included all the subsets of the French general population, although pediatrics were underrepresented. Prevalences of several illnesses (diabetes, asthma, atrial fibrillation, aortic aneurism), treatments (dyslipidemia, diabetes), patients’ profiles (dyslipidemia, atrial fibrillation, venous disease) were in agreement to those encountered in literature. However, smoking status, hospitalizations, referral to specialists were only partially reported and no information was available about sociodemographic status or death of patients. The availability of missing information through the use of questionnaires/pop up screens for physicians and patients, and the linkage of the EMR database to a claims database (HEAD) is also documented. CONCLUSIONS: We found no indications of lack of representativeness or validity of the LPD. While presenting some flaws associated with its naturalistic nature, LPD is a good support for pharmacoepidemiological and pharmacoeconomics studies.