In　this　paper,　a　comprehensive　error　analysis　method　is　proposed　to　discover　how　the　geometric　error　propagation　through　every　motion　axis,　and　to　find　out　which　error　parameters　have　greater　impact　on　the　tool　posture　error　at　the　end　of　the　kinematic　chain.　As　the　geometric　error　of　a　motion　axis　can　be　regarded　as　the　differential　movement,　an　error　model　for　a　four-axis　machine　tool　is　established　to　calculate　the　tool　posture　error　with　all　the　geometric　error　parameters.　Then　a　cumulative　process　of　the　differential　movements　of　every　axis　is　proposed　to　describe　the　error　propagation　process　when　moving　the　tool　to　the　given　position.　Moreover,　the　workspace　of　the　machine　tool　is　discretized　into　an　amount　of　points　with　a　uniform　sampling　method　on　the　measured　positions　of　the　geometric　error.　Then,　a　Spearman　rank　correlation　method　is　presented　to　find　out　how　closely　linked　between　a　single　error　parameter　and　the　tool　posture　error　all　over　the　sampling　workspace.　Hence,　the　ten　key　error　parameters　are　selected　according　to　the　analysis　results　in　the　three-axis　and　four-axis　sampling　workspace.　Finally,　an　experiment　is　conducted　on　the　four-axis　machine　tool　with　a　three-axis　controlled　trajectory　to　verify　the　effectiveness　and　correctness　of　the　proposed　method　using　a　double　ballbar.　(C)　2016　Elsevier　Ltd.　All　rights　reserved.