Various procedures and guidelines have been suggested for the development and construction of criterion-referenced tests. The present paper proposes a comprehensive model which allows the user to identify and relate specific components which affect the optimal construction and implementation strategies of criterion-referenced tests. Furthermore, it establishes parameter values which will allow the classification of individuals into mastery or nonmastery states with prespecified levels of confidence. Although the discussed model incorporates binomial expansions, it uses parameters of selected items for establishing baselines probabilities instead of true scores derived from an assumed population of items. (Author)