During the last 20 years, the relationship between reading instruction and assessment has been framed by the logic of mastery learning. The ultimate goal of mastery learning is assuring a given achievement outcome across students by varying input characteristics such as the amount and kind of instruction and practice, or the attention paid to prerequisite skills and aptitude characteristics of individual learners. Although the focus on testing has increased with the call for educational reform, currently available tests do not reflect advances in reading theory, practice, or research. The best possible assessment of reading seems to occur when teachers observe and interact with students as they read, evaluating the way in which the students orchestrate resources to construct meaning. An effective framework for complete reading assessment takes into consideration the relationship between objectives, decision-making units, and methods of assessment, and contains three critical features: (1) the attributes should reflect a theoretically sound model of the reading process, (2) the attributes should be highly interdependent and cannot be measured discretely, and (3) whatever is worthy of assessment ought to be assessable in different contexts for different purposes using a variety of strategies. In refocusing statewide assessment of reading, novel concepts that encourage strategic reading and that redefine the assessment-instruction link are being evaluated, such as summary writing, metacognitive judgments, question selection, multiple acceptable responses, and topic familiarity. The goal for reading instructors should be to develop valid, reliable, and usable strategies for assessment of reading comprehension. (Twenty-one references, a list of contrasts between new views of reading and current practices in reading assessment, and a framework of the relationship between objectives, decision-making units, and assessment methods are included.) (NKA)