The problem of using a domain-referenced system of achievement tests is discussed as it relates to the design of instruction. Testing problems are discussed from the point of view that the teacher, pupil, and/or automation needs certain kinds of information in order to make instructional decisions that are adaptive to the individual learner. The design of achievement tests based on item forms is determined by the purpose(s) for which the information obtained from them is needed. The selection of items from the defined domain of item forms is discussed in terms of the purpose for testing, the relationship between items and instruction, and the relationship between instructional objectives and item forms. (author)