The　key　problem　of　calculating　profile　error　for　a　gas　turbine　blade　is　discussed,　and　the　non-uniform　rational　B-splines　(NURBS)　surface　theory　is　adopted　to　describe　the　gas　turbine　blade.　The　profile　error　of　the　gas　turbine　blade　is　defined,　and　its　mathematic　model　is　established.　The　subdivision　approach　algorithm　is　used　to　calculate　the　minimum　distance　between　the　point　and　the　surface　so　that　the　speed　of　data　processing　can　be　enhanced.　The　combination　of　the　subdivision　approach　algorithm　and　the　particle　swarm　optimization　is　employed　to　realize　the　optimum　fitting　between　measurement　points　and　the　complex　surface　and　then　obtain　the　profile　error　of　the　blade　with　global　optimization.　This　data　processing　method　is　in　accordance　with　the　least　condition　principle,　and　it　can　be　applied　to　three-coordinate　measuring　machines.