This　paper　investigates　the　multiple　attribute　decision　making　(MADM)　problem　with　fuzzy　preference　information　on　alternatives　and　proposes　an　eigenvector　method　to　rank　them.　Three　optimization　models　are　introduced　to　assess　the　relative　importance　weights　of　attributes　in　a　MADM　problem,　which　integrate　subjective　fuzzy　preference　relations　and　objective　information　in　different　ways.　Since　the　decisions　these　models　lead　to　are　based　on　both　subjective　and　objective　information,　they　are　more　balanced　and　credible　than　those　that　consider　only　one　type　of　information.　An　example　is　provided　to　illustrate　the　application　of　the　proposed　models.　©　2005　Elsevier　B.V.　All　rights　reserved.