In this paper, we present the ER-MCDA methodology for multi-criteria decision-making based on imperfect information coming from more or less reliable and conflicting sources. The Analytic Hierarchy Process (AHP), Fuzzy Sets, Possibility and Belief Functions theories are combined to take a decision based on imprecise and uncertain evaluations of quantitative, qualitative criteria. Classical aggregation of criteria is replaced by a two-step fusion process using advanced fusion rules based on the Dezert-Smarandache Theory (DSmT) that allows to make a difference between importance, reliability and uncertainty of information sources and contents.