A reasoning mechanism capable of dealing with imprecise and uncertain information is essential for expert systems. In this paper, we propose the use of truth-qualified fuzzy propositions as the representation of imprecise and uncertain information, where