Item writing methods for criterion referenced tests in the cognitive domain are reviewed. The methods are classified on a continuum from informal-subjective to automated-objective. Examples of approaches to improving instructional objectives are given. Domain-based tests are derived from content specifications and five approaches are described: (1) item forms; (2) linguistic-based approaches; (3) facet theory; (4) concept-based testing; and (5) computer-based methods. Rules which govern item construction and which specify the structure, format, and some of the wording are called item forms. Linguistic-based items are based on directions to the item writer on how to rearrange segments of the instruction to obtain a certain item type. Facet theory is viewed as a way to develop a population of items representing the domain of instruction. Concept-based testing is used to test a student's understanding of a concept by checking for generalization to new examples and discrimination of non-examples. Computer systems for creating domain-based criterion referenced tests are available. Areas for improvement are described as the definition of domains and creation of item-writing algorithms, both at the conceptual and factual levels. Bibliographic citations are appended. (MH)