The　phenomenon　of　liquefaction　is　significant　during　the　Wenchuan　earthquake,　and　the　gravel　soil　liquefaction　has　accounted　for　a　large　proportion　of　it.　Due　to　the　high　dimensional　complex　nonlinear　relationship　between　the　gravel　soil　liquefaction　and　influencing　factors,　it　is　very　challenging　to　meet　the　requirements　of　the　speediness　and　high　precision　of　gravel　soil　liquefaction　due　to　the　intrinsic　defects　of　theoretical　analysis　and　numerical　calculation,　therefore,　a　new　prediction　method　based　on　Gaussian　Process　Regression　(GPR　model),　as　a　probabilistic　kernel　leaning　machine　and　powerful　tool　for　solving　highly　nonlinear　problems,　which　is　proposed　for　the　liquefaction　forecast　of　seismic　gravel　soil.　Six　measured　indicators　of　gravel　soil　liquefaction　were　selected　as　key　impacting　indicators　to　predict　the　main　impact　factor,　including　earthquake　intensity　EI,　the　peak　acceleration　of　earthquake　PGA,　the　depth　of　gravel　layer　ds,　groundwater　level　dw,　effective　overburden　pressure　σv＇,　and　shear　wave　velocity　Vs.　Firstly,　37　groups　typical　cases　were　collected　as　the　training　sample　and　8　groups　typical　cases　were　defined　as　the　test　sample,　which　has　enter　the　test　sample　and　output　the　prediction　probability　from　the　GPR　model.　Secondly,　the　SVM　method,　PLS　method,　MLR　method,　LSSVM　method　and　RF　method　were　chosen　to　verify　the　validity　and　reliability　of　GPR　model.　Finally,　a　example　analysis　show　that　the　method　is　feasible,　effective　and　simple　to　implement,　which　also　provide　a　new　way　for　the　liquefaction　prediction　of　seismic　gravel　soil.　Copyright　©　2015　Binary　Information　Press.