The assumptions of the Criterion-Referenced Test (CRT) model proposed by Kriewall are compared to those of Emrick and Adam's Mastery-Learning (ML) model. Testing, in the context of instructional management, serves three general purposes: performance evaluation (achievement of objectives), placement (classification of students for instruction), and diagnosis of learning deficiencies. Both of the test models discussed here assess the achievement of objectives; they differ in the types of objectives for which they are best suited. Both test models have potential usefulness for making placement decisions, but only the ML model is likely to be useful in diagnosing learning deficiencies. The applicability of each model for instructional management decisions is discussed. (Author/DEP)