A　fast　particle-locating　method　is　proposed　for　the　hybrid　Euler-Lagrangian　models　on　the　arbitrary　polyhedral　mesh,　which　is　of　essential　importance　to　improve　the　computational　efficiency　by　searching　the　host　cells　for　the　tracked　particles　very　efficiently.　A　background　grid,　i.e.,　a　uniform　Cartesian　grid　with　a　grid　spacing　much　smaller　than　computational　mesh,　is　constructed　over　the　whole　computational　domain.　The　many-to-many　mapping　relation　between　the　computational　mesh　and　the　background　grid　is　then　specified　through　a　recursive　tetrahedron　neighbor　searching　procedure,　after　the　tetrahedral　decomposition　of　computational　cells　and　a　mapping　inverse　operation.　Finally,　the　host　cell　is　straightforwardly　identified　by　the　point-in-cell　test　among　the　optional　elements　determined　based　on　the　mapping　relation.　The　proposed　method　is　checked　on　three　meshes　with　different　types　of　the　cells　and　compared　with　the　existing　methods　in　the　literatures.　The　results　reveal　that　the　present　method　is　highly　efficient　and　easy　to　implement　on　the　arbitrary　polyhedral　mesh.　©　2019　by　the　authors.