The　S-shaped　test　piece　is　utilized　to　comprehensively　detect　the　machining　accuracy　of　multi-axis　machine　tools.　However,　due　to　the　existence　of　the　influence　of　theoretical　error　for　S-shaped　test　piece　on　the　acceptance　of　multi-axis　machine　tools,　therefore,　a　new　methodology　to　reduce　the　theoretical　error　is　proposed　in　this　study.　Firstly,　the　uniform　double　cubic　B-spline　surface　model　is　applied　to　a　surface　representation.　By　utilizing　this　model,　a　model　of　the　S-shaped　test　piece　is　established.　Furthermore,　the　distribution　of　the　twist　angle　at　the　selected　three　section　lines　is　obtained.　Meanwhile,　the　influence　of　twist　angle　on　the　theoretical　error　of　S-shaped　test　pieces　is　analyzed.　Then,　the　single-point　offset　(SPO)　method　is　analyzed,　which　has　some　drawbacks.　In　order　to　significantly　reduce　the　theoretical　error　of　S-shaped　test　piece,　the　optimized　single-point　offset　(OSPO)　method　is　proposed.　Moreover,　the　tool　path　of　the　S-shaped　test　piece　is　generated　by　CAD/CAM　software　based　on　the　OSPO　method.　Finally,　in　order　to　verify　the　feasibility　of　the　presented　method,　a　machining　and　measurement　experiment　is　carried　out　on　the　gantry-type　five-axis　milling　machine　tool　(XKAS2525)　and　coordinate　measuring　machine　(CMM)　based　on　the　S-shaped　test　piece,　respectively.　Experiment　results　show　that　the　average　theoretical　error　of　S-shaped　test　pieces　based　on　the　OSPO　method　is　reduced　by　about　50.1%　than　that　on　the　SPO　method.　In　addition,　the　vast　majority　of　theoretical　errors　based　on　the　OSPO　method　are　approximately　less　than　5　mu　m,　which　can　be　negligible.　It　is　therefore　reasonable　to　conclude　that,　compared　with　the　SPO　method,　the　proposed　method　in　this　paper　can　avoid　the　influence　of　theoretical　error　on　the　acceptance　of　multi-axis　machine　tools　intuitively　and　efficiently.