Attribute　reduction　has　become　an　important　pre-processing　step　to　reduce　the　complexity　of　data　mining.　Rough　set　reduction　and　correlation-based　methods　have　gradually　contribute　towards　improving　attribute　reduction　techniques.　Many　researchers　have　proven　that　the　rough　set　reduction　method　is　effective　in　reducing　redundant　attributes　without　information　loss.　Correlation-based　methods　evaluate　attribute　as　a　subset　reduce　irrelevant　instead　of　individual　attribute.　In　this　paper,　we　propose　a　new　method　(RSCBA)　of　combing　correlation-based　methods　and　rough　set　to　reduce　irrelevant　and　redundant　attributes　in　a　more　effective　way.　The　UCI　datasets　were　used　to　verify　the　effectiveness　of　the　proposed　method　compared　to　the　rough　set　(RS)　method　and　the　correlation-based　feature　selection　(CFS)　method.　Experimental　results　show　that　our　method　obtains　comparatively　higher　reduction　strength　and　classification　accuracy.