A randomized trial involving 19 elementary schools (K-5) was conducted to replicate and extend two previous experimental studies of the effects of a voluntary summer reading program that provided (a) books matched to students' reading levels and interests and (b) teacher scaffolding in the form of end-of-year comprehension lessons. Matched schools were randomly assigned to implement one of two lesson types. Within schools, students were randomly assigned to a control condition or one of two treatment conditions: a basic treatment condition replicating procedures used in the previous studies or an enhanced treatment condition that added teacher calls in the summer. During summer vacation, students in the treatment conditions received two lesson books and eight books matched to their reading level and interests. Overall, there were no significant treatment effects, and treatment effects did not differ across lesson type. However, there was a significant interaction between the treatment conditions and poverty measured at the school level. The effects of the treatments were positive for high-poverty schools (Cohen's d = 0.08 and 0.11, respectively), defined as schools where 75-100% of the students were receiving free or reduced-price lunch (FRL). For moderate-poverty schools (45-74% FRL), the effects of the treatments were negative (Cohen's d = -0.11 and -0.12, respectively). The results underscore the importance of looking at patterns of treatment effects across different contexts, settings, and populations.