This paper provides a review of a class of probabilistic models that has been developed for use in the assessment of trait or competency acquisition. Consideration is given to the relative merits and limitations of this class of state models, under which trait acquisition is conceived as being "all-ornone," as compared with those occurring under an alternative conceptual framework, in which trait acquisition is assumed to be gradual. In addition, some of the applications of these state models are presented, including the establishment of mastery classification decisions and the assessment of consistency with respect to items and classification. Finally, some extensions to the class of state models, which may be helpful in increasing the applicability of this class of models, are presented. An important element of the criterion-referenced approach to testing is the assessment of individuals in terms of absolute standards of attainment for traits or competencies of interest; and within this context, the concept of &dquo;mastery&dquo; has played a central role. To deal with this classification problem, a variety of strategies have been suggested (see Hambleton & Eignor, 1979). Meskauskas (1976) pointed out that most of these strategies can be grouped into two general classes based on the underlying conceptualization of the trait being assessed. Within the first class, called continuum models, trait acquisition is assumed to be gradual and mastery is viewed as an interval on a test score scale. Within the second class, called state models, trait acquisition is conceived of as an &dquo;all-or-none&dquo; process and mastery is viewed as the presence of trait acquisition. Limitatlons of Continuum Models From the perspective of continuum models, mastery of a trait is based on &dquo;sufficient&dquo; partial acquisition of that trait. Thus, mastery classification involves a judgmentally established point or set of points on the trait continuum. An integral part of procedures used to establish rules for mastery classification within the framework of continuum models, mentioned by both Linn (1978) and Glass (1978), is their fundamentally judgmental nature, resulting in rules for classification that are arbitrary. This arbitrariness presents a problem for effective implementation of the procedures; and as might be expected, consistency of mastery classifications both within and across procedures often is