This work addresses the problem of medical terminologies building. Starting from a set of elementary knowledge, a multi-hierarchical terminology is generated according to formal principles that are based on explicit classificatory points of view. The experimentation has been carried out on kidney diseases. Hierarchical conceptual structures are generated starting from Conceptual Graph based definition of kidney disease, a hierarchy of defining types and a set of organizing rules. The generated hierarchy includes new potential disease concepts with different level of likeliness. This “ready to use” generated multi-hierarchical terminology can be used for coding, includes new potential terms for the future and, due to the formalization of classificatory principles, offers improved opportunities of diseases instances aggregation. These features are looking very promising in the context of epidemiological observation tools (registries, long term cohort studies). They offer mean to meet two critical requirements: the statistical continuity in data analysis and the sustainable management of the underlying terminological system.