Standardized reading tests exert a great influence on curriculum materials, teaching methods, and student perceptions of the nature of reading; these tests must reflect the best recent research findings. Passage dependency research has attempted to increase the accuracy of reading tests, and has focused on the following four areas: (1) reading comprehension as the amount of knowledge the reader gains from the reading, (2) reading comprehension measured by students' ability to answer questions after having read the passage that they could not have answered beforehand, (3) identification of strategies students use in answering questions correctly without the aid of the passage, and (4) recommendations for making reading comprehension tests more passage dependent. There is a need to go beyond such research to a psycholinguistic definition of reading comprehension in which comprehension is the construction of meaning, not merely a passive discovery of meaning in text. Increasing the dependency of the reader on the passage might very well inhibit worthwhile information processing strategies. As opposed to the "manipulative" techniques of much of the mainstream research in this area, Kenneth Goodman's naturalistic model of reading is a valid alternative, working within real reading situations and emphasizing miscue analysis to illumine the comprehension process. (DF)