This study examined the role of skills analysis in curriculum-based measurement (CBM) for the purpose of developing more effective instructional programs. Participants were 30 special educators who were assigned randomly to three treatment groups: (a) CBM with graphed performance indicator and skills analysis, (b) CBM with graphed performance indicator analysis only, and (c) no systematic performance monitoring (control). Teachers implemented their respective treatments for 15 weeks with a total of 91 mildly and moderately handicapped pupils. Analyses of variance indicated that use of CBM with both performance indicator and skills analysis resulted in more specific instructional changes and better student achievement than CBM with performance indicator analysis alone or control. Implications are discussed for CBM practice within special education and for the more general field of educational assessment.