For　the　gears　utilized　in　wind　turbines,　the　main　reason　leading　to　malfunction　is　the　crack　initiation　and　propagation　under　cyclic　loading.　In　the　gear　engagement　process,　the　fatigue　damage　accumulates　gradually　and　formulate　visible　microcrack　if　the　fatigue　parameter　reaches　to　specified　threshold.　In　order　to　simulate　the　fatigue　evolution　in　wind　turbine　gears,　a　mesoscopic　model　is　established,　in　which　the　Voronoi　tessellation　and　cohesive　elements　are　integrated,　such　that　the　crystal　structure　and　grain　boundary　can　be　modeled.　The　dy-namic　stress　distribution　and　fatigue　evolution　phenomenon　are　then　investigated　under　different　working　conditions.　The　loading　in　the　meshing　region　is　simplified　as　moving　load　on　a　semi-infinite　plane,　so　the　shear　stress　distributions　for　per　loading　cycle　are　obtained.　Furthermore,　bilinear　traction-separation　law　(TSL),　Jump　-in-Cycles　(JIC)　loading　methodology　and　multi-axial　fatigue　criterion　are　involved　to　fulfill　the　damage　accu-mulation　simulation.　As　a　result,　the　initiation　location　and　propagation　pattern　of　the　fatigue　crack　are　pre-dicted,　and　the　fatigue　life　curve　is　also　obtained.　Thereafter,　the　influence　of　wind　speed　and　lubrication　condition　on　fatigue　damage　evolution　is　also　discussed.　The　results　have　shown　that　the　wind　loading　mainly　influences　the　magnitude　of　the　maximum　shear　stress,　whereas　the　contact　width　will　determine　the　location　of　the　maximum　shear　stress.　When　the　whole　fatigue　evolution　procedure　is　finished,　the　first　two　cracks　initiate　at　the　depth　of　0.13　mm　and　0.27　mm　beneath　the　surface　respectively,　and　then　propagate　into　the　substrate.　Through　comparative　analyses　of　fatigue　life　curves,　it　can　be　concluded　that　higher　wind　speeds　and　poor　lubrication　conditions　will　reduce　gear　service　life　significantly.