Abstract: in 　the 　stud ywe　combined　continuum　removalcr with　a　partial 　least 　squares 　re gressionplsr method　to　build　inversion　models　of 　soil　as　concentration　and　or ganic　matteromcontentbased　on　hypers pectral 　data　between　400－2　500nm　of 　100soil 　samples　which　were　collected 　in　yixingre gion 　of 　jian gsu　province．the 　results 　show　thatcompared　with　the 　common　plsr　modelsmodel＇sdetermination 　coefficientsr2and　root　mean　square　error 　of 　prediction rmsepare 　0．5123．0９0and0．6215．９34res pectivel y cr－plsr　models　present 　an 　improvement 　of　as　and　om　inversion　models r2　and　rmsep　are 　0．7632．323and　0．９114．5９９res pectivel y．in　addition cr　effectivel ystren gthens 　the 　spectral 　characteristics　of 　550９0014201９00and　2　200nm　wavelengths．the 　anal ysis 　of 　re gression 　coefficients 　of 　cr－plsr　and　plsr　models　of　as　and　om　demonstrate 　several 　important　wavelengths 　enhanced　by　cr　and　are 　simultaneousl y　used 　for　cr－plsr　models　of　as　and　om．thus　cr　is 　able 　to　hel p　plsr　conduct　wavelength　selection　and　enhances 　the 　inversion　precision　ofsoil　as　and　om　content 　effectivel y．this　findin g　can　be　used　as 　a　reference 　for 　remote 　sensing　monitorin g　of 　soil 　qualit y．