For　reducing　the　fault　ratio　of　shaft　ore-roasting　furnace,　based　on　the　analysis　of　the　fault　mechanism　and　combination　of　case-based　reasoning　(CBR)　and　variables　prediction,　an　intelligent　fault　prediction　approach　is　proposed　for　the　shaft　furnace　roasting　process.　The　prediction　model　of　the　process　variables　performs　to　predict　key　technical　parameters　as　the　fault　symptoms　that　is　hard　to　measure　online.　The　probability　of　the　typical　fault　and　their　operation　guidance　with　the　help　of　case-based　reasoning　technology　are　obtained.　The　proposed　fault　prediction　system　is　successfully　applied　to　the　roasting　process　of　a　shaft　furnace,　the　fault　ratios　during　production　process　is　decreased,　and　the　proved　benefit　is　achieved.