The surveillance of chronic diseases in France relies on various data sources: general population surveys, registries, medical causes of death, and medical-administrative databases (BDMA). These databases, which are comprehensive and responsive—though not originally designed for epidemiological purposes—now serve as an additional data source. The objective is to describe the BDMA, highlighting their strengths, limitations, and complementarities relative to other data sources, through two examples: the surveillance of diabetes and cancers. MDA databases cover all reimbursed outpatient care and hospital stays for the population residing in France. They enable the identification of diseases or disease processes through their treatment patterns, the evaluation of prevention or screening initiatives, and more. The comprehensiveness of these databases allows for breakdowns at fine geographic levels, and their timeliness provides data for the most recent years. BDMA databases can serve to control for recall bias when cross-referenced with individual questionnaire surveys, or serve as a basis for adjustment by comparing respondents and non-respondents (as in the example of diabetes). However, since they are based on reimbursed care, they evolve with treatments, reimbursement policies, etc. Their performance as an epidemiological surveillance tool must be regularly reassessed. Several evaluation methods are possible depending on the objectives and other existing data sources. Health insurance databases can contribute to the surveillance of chronic diseases, especially as their quality—in terms of both coding and content—continues to improve, opportunities for cross-referencing with other data sources increase, and conditions for their accessibility become clearer. (R.A.)