A distinction is made between the learnability and readability of text materials. Learnability refers to the extent to which new learning results from reading a passage; readability refers to the extent to which a passage is comprehended. Cleraly, comprehension can occur without new learning. Classic readability formulas use text characteristics such as word difficulty and sentence length to predict comprehension. In this study, the prediction of new learning with a variety of text variables was examined. Sixty-two sixth-grade subjects read 11 passages ranging from second- to twelfth-grade difficulty. Learning scores were based on two types of pretests and post-tests. The results of stepwise regression analyses of text characteristics on learning scores revealed that the same characteristics, including word difficulty and sentence length, which predict comprehension also predict new learning. However, more than one text characteristic in a multiple regression equation did not significantly improve the predictability over the zero order correlation. Thus, classic readability formulas may be used to estimate the learnability as well as the readability of text materials. (Author/NH)