The paper investigates and analyses the current state of the art of criterion-referenced measurement (CRM), with a view to determining its use in training and instructional programs. It presents a reveiw of the literature pertaining to the following aspects: a brief history of CRM; a definition and comparison of criterion-referenced and norm-referenced measures; usage of the two measures; and the construction and evaluation of criterion-referenced tests in terms of validity, reliability, and other test characteristics. The literature supports the following conclusions: (1) all definitions of CRM stress score interpretation as representing what the individual can do relative to instructional objectives rather than other individuals; (2) criterion-referenced information is valuable in making certain decisions based on what a person can do at a given point in the training cycle; (3) CRM has focused much attention on behavioral objectives and training outcomes; (4) behavioral objectives must be carefully written to effectively direct and measure instruction; (5) more than one measure should be used to validate any CRM to decrease the error associated with its measurement; (6) it is difficult to develop objective procedures necessary for CRM of complex behavior; (7) CRM supplements but should not replace normative tests in training; and (8) more research is needed before extensive use of CRM in instructional programs can be recommended. (Author/NJ)