The　accurate　determination　of　the　level　of　spontaneous　combustion　tendency　of　sulfide　ores　can;　provide　important　theoretical　basis　for　the　design　of　sulfur-rich　mine　fire　prevention　and　safety　mining.　In　this　paper,　based　on　the　existing　research　results,　three　main　indexes　that　most　reflect　spontaneous　combustion　tendency　of　sulfide　ores　were　taken　into　account　to　regard　as　basic　discriminant　factors　of　the　classification　model,　including　the　oxidation　increment　rate　of　ore　sample　under　low　temperature　in　laboratory,　self-heating　temperature　and　autogenous　ignition　temperature.　Eighteen　groups　of　samples　collected　from　typical　sulfur-rich　mine　was　taken　as　training　samples　for　analysis　and　calculation,　then　the　GA-BP　neural　network　model　of　spontaneous　combustion　tendency　classification　of　sulfide　ore　was　established.　Finally,　by　the　discriminant　model　of　spontaneous　combustion　tendency　of　sulfide　ore　samples,　six　groups　of　sample　from　typical　mines　were　discriminated.　The　results　of　this　study　show　that　the　misclassification　rate　average　for　six　groups　of　measured　samples　is　only　2.5%,　and　achieve　good　prediction　effect.　Therefore,　GA-BP　neural　network　model　can　be　used　for　guiding　the　classification　of　spontaneous　combustion　tendency　rating　for　sulfur-rich　ores.