California and Texas mandate 75% to 80% decodable texts for first-grade reading programs, yet these percentages have no empirical base. This study examines the text selections in 6 first-grade programs from the perspective of lexical, semantic, and syntactic features. The composition of text differed across the 6 programs with respect to length, grammatical complexity, the number of unique and total words, repetition of words, and coverage of important vocabulary. Potential decoding accuracy rates, as computed by a relational database program created for this project, varied widely across the 6 programs and often depended heavily on holistically taught words. Moreover, the majority of words appeared only once in each 6-week instructional block across the year. Implications are discussed for (a) learning to read from basals, (b) design of basal reading programs, and (c) future research.