Identification　of　the　spindle　unbalance　is　very　important　for　the　ultra-precision　machining　tools.　In　the　first　part,　the　frequency　information　for　the　spindle　system　is　computed　by　two　methods.　In　the　second　part,　the　results　between　the　modal　information　of　the　spindle　and　the　error　frequency　of　the　measured　workpiece　surface　which　processed　by　wavelet　transform　and　power　spectral　density　is　compared,　the　signal　feature　in　the　waviness　which　is　consistent　with　the　spindle　unbalance　frequency　is　extracted.　In　the　last　part,　the　identified　result　is　compared,　it　proves　the　extraction　and　identification　method　for　the　spindle　unbalance　of　the　machine　tool　is　correctness　and　validity.　(c)　2012　Elsevier　Ltd.　All　rights　reserved.