This study estimates program effects (English immersion versus primary language instruction using varying degrees of Spanish) upon English and Spanish word identification in the context of changing classrooms (cross-classification) from kindergarten through second grade. Letter-word identification is an important predictor of early reading achievement (Scarborough, 2001, 2005). In particular, the authors are interested in classroom and school differences not merely as noise to be removed, but as indicators of substantively important differences in implementation and practice. The student participants were 1,991 children (49% female) taught by 421 teachers across the 3 years of the longitudinal project, kindergarten to second grade. Seventy-five percent of the families reported earning less than $30,000 per year, and 79% had foreign-born fathers. Results suggest that in English, Spanish language instruction may facilitate the learning of English over time, as evidenced by the maintenance programs catching up to the English performance of the immersion programs. In Spanish letter-word identification, immersion programs show improving performance over time, but at a slower pace than the Spanish maintenance programs. These program-by-time interactions, however, are not completely correct and will require further verification. (Contains 4 figures and 3 tables.)