Among the advantages of a well-written review of the reading research literature are that it lays down the framework of the topic for the reader and provides exercise for both the synthesizer and the reader in integrating widely scattered, incompatible research findings. These benefits are intimately related to the problems in such syntheses, such as the limitations and loopholes of a computer data base search and the fallibility of the researcher's judgment, selection, and classification systems. Much writing and rewriting of the research review are required to ensure clarity, sustained interest, and balance. Most of what takes place in developing a research synthesis is descriptive, rather than prescriptive. Research synthesis, however, should play a prominent role in the prescriptive domain in order to provide both the descriptions of reading behavior and the design for changing that behavior through educational practice. Recommendations based on such syntheses should generate more confidence in their scientific underpinnings than do single studies. Such reviews appear infrequently in the reading research literature because few students receive systematic training in conducting research syntheses and because such work tends to be undervalued in comparison with "pure" research. (HTH)