In　the　state　of　machining　service,　to　evaluate　the　machining　accuracy　reliability　of　multi-axis　CNC　machine　tools,　analyze　the　accuracy　failure　mode　and　the　reliability　sensitivity　under　the　failure　mode,　this　paper　proposes　a　research　method　by　the　cross-correlation　studies　of　geometric　error　parameters　to　improve　and　promote　the　accuracy　reliability　of　CNC　machine　tools.　Firstly,　by　multi-body　system　theory,　the　homogeneous　coordinate　transformation　matrix　between　each　body　of　the　machine　tool　is　established,　and　the　spatial　machining　accuracy　model　is　constructed.　At　the　same　time,　considering　the　time　series　problem　in　the　measurement　process　of　geometric　error　parameters,　this　paper　studies　the　correlation　between　the　error　parameters　for　reliability　analysis　and　proposes　a　novel　reliability　analysis　index　for　each　measurement　point　in　the　machine　tool　workspace.　To　validation　of　the　method,　the　analysis　results　of　this　paper　are　compared　with　those　by　Monte　Carlo　simulation.　In　addition,　the　accuracy　failure　conditions　of　the　machine　tool　are　studied,　the　failure　state　function　of　machine　tool　accuracy　is　established,　and　the　accuracy　failure　modes　that　may　occur　are　analyzed　from　this,　to　carry　out　the　accuracy　failure　sensitivity　analysis　under　the　failure　mode,　and　to　identify　key　geometric　errors　which　have　a　great　impact　on　the　machining　accuracy　reliability　of　machine　tool　in　the　failure　mode.　Finally,　taking　a　3-axis　machine　tool　as　an　example,　according　to　the　analysis　results,　this　paper　puts　forward　measures　to　improve　the　accuracy　reliability　and　verifies　the　feasibility　of　the　proposed　method.