This　paper　proposes　an　efficient　and　automated　scheme　to　predict　and　identify　the　position　and　motion　errors　of　rotary　axes　on　a　non-orthogonal　five-axis　machining　centre　using　the　double　ball　bar　(DBB)　system.　Based　on　the　Denavit-Hartenberg　theory,　a　motion　deviations　model　for　the　tilting　rotary　axis　B　and　rotary　C　of　a　non-orthogonal　five-axis　NC　machine　tool　is　established,　which　considers　tilting　rotary　axis　B　and　rotary　C　static　deviations　and　dynamic　deviations　that　total　24.　After　analysing　the　mathematical　expression　of　the　motion　deviations　model,　the　QC20　double　ball　bar　(DBB)　from　the　Renishaw　Company　is　used　to　measure　and　identify　the　motion　errors　of　rotary　axes　B　and　C,　and　a　measurement　scheme　is　designed.　With　the　measured　results,　the　24　geometric　deviations　of　rotary　axes　B　and　C　can　be　identified　intuitively　and　efficiently.　This　method　provides　a　reference　for　the　error　identification　of　the　non-orthogonal　five-axis　NC　machine　tool.　(C)　2015　Elsevier　Ltd.　All　rights　reserved.