Present definitions of the criterion-referenced test are discussed, insufficiences noted, and a new definition proposed. Some examples of criterion-referenced tests are examined and used to educe some general principles for the development of such tests. The utility of item form processes is assessed. It is suggested that the difficulty of objectively defining a test construction process is directly proportional to the complexity of the behavior the test is designed to assess. Problems and doubts with regard to the development of criterion-referenced tests for complex behavior domains are noted. In addition, some empirical methods for dealing with item analysis, test reliability, and test validity difficulties are advanced. (DG)