In the assessment of second/foreign language proficiency, we are entering the era of criterion-referenced assessment as language learning is being recognized as an integrative, multifaceted construct. Norm-referenced measurement (NRM) is compared with criterion-referenced measurement (CRM). CRM is characterized by attention to skill, whereas NRM focuses on student rank. The evolution of some modern multi-componential language ability models is traced, starting with that of Canale and Swain (1980). CRM, with its greater focus on skills, should provide a better perspective from which to measure such a wide array of skills than NRM. One process for measuring skills is criterion-referenced language test development (CRLTD). CRLTD is characterized by flexibility. Participants iterate between the test planning and test item/task writing, cycling between test specification and product. CRLTD is a bottom-up group-based test development process. Although time constraints do not permit a thorough exploration of the technique in this paper, audience members are encouraged to try it to improve skills-based testing in the modern era of language teaching. Six figures illustrate the discussion. (Contains 8 references.) (SLD)