The　sensitivity　analysis　of　factors　that　influence　buried　pipeline＇s　seismic　behavior　may　help　us　find　out　which　one　is　dominant,　so　more　scientific　and　special　earthquake-resistant　measures　may　be　provided,　and　this　will　be　important　for　improving　the　buried　pipeline＇s　seismic　behavior.　In　this　paper　the　buried　pipeline＇s　maximum　axial　stress　under　the　earthquake　action　was　selected　as　the　index　which　represents　the　buried　pipeline＇s　seismic　behavior,　and　other　8　factors　were　selected　for　sensitivity　analysis　through　historical　earthquake　damage　experience.　Firstly,　the　prediction　model　for　buried　pipeline＇s　response　under　earthquake　action　was　established　based　on　General　Regression　Neural　Network　(GRNN).　Secondly,　the　sensitivity　analysis　of　the　above　factors　was　carried　out　through　combining　the　orthogonal　table　with　theory　of　utility　function　and　ANN,　at　the　same　time　the　assessment　model　was　set　up.　Finally,　the　application　results　was　compared　with　the　results　obtained　by　traditional　method,　and　its　reliability　and　practicability　were　proved.