A　simple　yet　pragmatic　method　of　consistency　test　is　developed　to　check　whether　an　interval　comparison　matrix　is　consistent　or　not.　Linear　programming　method　is　used　to　derive　consistent　interval　weights　from　consistent　interval　comparison　matrices　and　to　aggregate　local　interval　weights　to　generate　global　interval　weights.　In　cases　where　an　interval　comparison　matrix　is　inconsistent,　an　eigenvector　method-based　nonlinear　programming　(NLP)　approach　is　developed　to　generate　interval　weights　that　can　meet　pre-determined　consistency　requirements.　A　simple　and　effective　preference　ranking　method　is　utilized　to　compare　the　interval　weights　of　criteria　or　rank　alternatives.　Three　numerical　examples　including　a　hierarchical　(AHP)　decision　problem　are　provided　to　illustrate　the　validity　and　practicality　of　the　proposed　methods.　©　2004　Elsevier　Inc.　All　rights　reserved.