Clinical-administrative databanks are a key tool in support of public health decision-making. A number of databanks are available relevant to population needs, resources available, as well as performance indicators. Since the 2000s, considerable efforts have been dedicated to the consolidation of findings and development of tools aimed at improving surveillance with respect to the health status of populations and performance of the social and healthcare system. At the annual congress of the Association francophone pour le savoir (ACFAS), held in 2017 at McGill University, a seminar was organized on the utilization of databanks in mental health and in addiction. This seminar featured an expert discussion on subjects related to: identification of the principal clinical-administrative databanks, the extent of their use, their limitations, and solutions aimed at optimizing the development of databanks to better support the management of services. This article summarizes the content of this seminar. While databanks entail important strengths, including great potential for the generalization of information, they also present limitations regarding their capacity to respond to needs, quality and validation issues, as well as accessibility. Various recommendations were proposed to improve the management of databanks and optimize their impact, including their centralization in a single, and highly accessible autonomous organism, and societal and cultural change favoring performance evaluation in the interest of improving practices and better monitoring health results.