A guide for developing educational criterion measures is presented. Subjective evaluations are at best highly variable and conflicting. Specification of objectives precedes selection of criterion measures. The criterion measures presented are intended for evaluation of programs, not pupils. Frequently multiple criterion measures are necessary to effectively evaluate a given program. Criterion measures can be derived from either learner behavior or learner products such as test results. Conditions under which these measures are observed may be either natural or manipulated conditions, the latter being common. Absolute and relative criteria each serve separate functions. Absolute criteria rate performance in terms of specific instructional objectives. Relative criteria compare an individual's performance to that of the other pupils. All of these variables in criteria may be combined in different ways to produce the most valid and effective measures for evaluating a particular program. (SM)