A large field of applications is opened by introducing incomplete, vague or indefinite facts in data models. The fuzzy logic was applied to models and data bases. Extensions of relational interrogation languages to fuzzy logic were realized. When analyzing and assessing the commercial partners of an enterprise, an important aspect is grouping them in function of their common features. This regrouping in classes contributes to reducing complexity and allows an improved analysis of commercial relationships with partners. In a fuzzy classification, each partner could belong, upon the attributes values, to one or more classes. For example, a partner could simultaneously belong to the class of “partners with doubtful quality” (with an affiliation value of 0.3) and to the class of “partners with increasing perspectives of commercial cooperation” (with a value of 0.7). Fuzzy classification thus allows a differentiated interpretation of the affiliation to various classes. This paper proposes a model of a fuzzy relational data base for managing the relationship with suppliers. The advantages of a fuzzy classification are thus emphasized. Classification requests in FCQL (fuzzy Classification Query Language) language use linguistic variables in a fuzzy context.