In　this　paper,　a　method　of　complement　of　fuzzy　rough　set　and　BP　neural　network　was　proposed,　and　an　early　warning　model　of　electronic　information　products　on　Technical　Barriers　to　Trade　(TBT)　was　given　by　the　method.　The　attribute　reduction　for　indicators　of　early　warning　based　on　fuzzy　rough　set　can　not　only　enhance　the　veracity　of　attribute　reduction,　but　also　improve　the　accuracy　of　the　training　of　BP　neural　network　through　reducing　the　input　dimension　of　BP　neural　network　at　the　same　time.　The　new　TBT　early　warning　model　of　electronic　information　products　was　proved　more　feasible　and　effective.　©2010　IEEE.