This paper reviews the literature related to criterion-referenced testing (CRT), which has been tied to trends in individualized instruction, competency-based education, and the accountability movement. Among the conclusions that are drawn are (1) CRT scores assume a good match between test items and behavioral objectives that the CRT attempts to measure, and, to the extent that the test items do not accurately measure the objectives, the results will be inaccurate; (2) the theory on determining mastery levels is rather fragmentary and inconclusive; (3) reliability and validity studies on CRT are much needed; (4) CRT measurement should be based on a threshold loss function rather than a squared-error loss function; (5) Bayesian statistical procedures should be used to improve proportion-correct estimates; and (6) customized criterion-referenced tests need to be developed by local school districts by contracting with testing organizations. It is concluded that there is a need for awareness of both the strengths and weaknesses of criterion-referenced testing. (AA)