The question considered in this essay is: Given what we know about research and development in education, what might reasonably be expected from a National Institute of Education (NIE)? Two suggestions are made. One way to figure out what an NIE might be good for is to make up a list of what's wrong with American education and dream up some plausible ways in which R&D might help. A more cautious way of proceeding might be to set out what we know about education and R&D in advance and try to infer from that what might reasonably be expected from a government R&D agency. Subjects discussed include: educational problems, defining goals for an NIE, the knowledge base, utilizing R&D results, and definition of research and development. Consequences for NIE include: (1) Accepting a"major problems" mandate is not likely to reduce the potential for instability in an NIE; (2) Tendency toward political instability will be compounded by the fact that evaluative criteria for educational R&D are murky; and (3) There is a formidable system of constraints and there's no simple way to avoid the problems they pose. (For related document, see AA 001 048.) (CK)