A case is made for combining correlational and experimental approaches in research of expository prose understanding. Both approaches yield similar results and an ideal database is available in the form of multiple-choice tests. Influence of this kind of research on theories of prose comprehension and implications for future research are listed.