Since mid-1971 the Instructional Objectives Exchange has been engaged in a major effort to develop and disseminate criterion-referenced tests in the fields of reading, mathematics, language arts, and social studies. This paper isolates the chief technical decision-alternatives faced in this project, such as: (1) the optimal number of tests to produce, (2) choosing a well defined behavior domain, (3) devising a homogeneous domain, and (4) the high cost of preparing criterion-referenced tests. The author describes the rationale for each decision, then appraises the adequacy of the decisions on the basis of empirical results. (Author/MLP)