A series of studies on conceptual and design problems in competency-based measurements are explained. The concept of validity within the context of criterion-referenced measurement is reviewed. The authors believe validation should be viewed as a process rather than an end product. It is the process of marshalling evidence to support interpretations. Performance standards are important in the applications of criterion-referenced measurement. Shepard suggested standard setting should be an iterative process and one that involves various audiences. The problem of deciding on test length and passing score required to make sufficiently accurate mastery decisions is discussed. Self scoring, or computer administered, tests might provide savings in average testing time by making decisions sequentially. Item analyses for criterion referenced measures are likely to be as relevant to investigations of instructional processes as they are to concerns about item characteristics. In principle, item analysis results should not be used for item selection. Reliability procedures based upon assessment of consistency of decisions appeal to the authors as possessing the greatest utility. Estimation of domain scores, test bias, and matrix sampling are reviewed at length. (DWH)