Noting that accurate procedures for predicting reading ability in school settings are necessary for a variety of purposes, this paper reviews methods for predicting reading ability and presents a new, more precise model for use at the second grade level. First, the paper defines reading expectancy as an informed estimate of the level of reading achievement that can realistically be expected of an individual when the relevant facts are known. Presented next is a historical review of reading expectancy formulas from the late 1920s to the present. The review notes that many of the formulas grew more complex and began to test different aspects of reading, such as listening comprehension, while others relied on general intelligence testing to predict reading ability. After pointing out the difficulty these models have in predicting actual achievement for individual children, the paper proposes a multiple regression model that makes use of stepwise regression analyses to test second grade native speakers of English. Correlates of reading achievement that the proposed model must account for are the topic of the next section, which discusses intelligence tests, listening comprehension tests, anxiety, attention span, self-esteem, socioeconomic status, and sex variables. Twelve tests for these variables for inclusion in a proposed reading diagnosis battery are then listed, followed by three research hypotheses and a graph of the proposed reading expectancy model. The document concludes with 67 references. (SKC)