Current national efforts to expand the role of educational assessment and radically change the nature of assessment are discussed. Rationales for expectations that a new national examination system would serve as a lever of educational reform are analyzed. Challenges posed for educational measurement by the proposed heavy reliance on complex, performance-based assessment are examined in terms of the needed comprehensive validation research. Particular attention is given to existing generalizability evidence and the implications of high degree of task specificity in performance for the design of an assessment system. Nine figures and one table illustrate the discussion. (Contains 45 references.) (Author)