It is commonly believed that reading challenges should be addressed early to reduce the likelihood that developmental delays will impact students over the long term. However, students in developing countries often have limited access to reading resources. This study uses a randomized controlled trial to examine the causal effect of a classroom library program on student reading outcomes and academic achievement in schools with poor reading resources over an eight-month period. We find that the program significantly improved student reading habits (borrowing books, reading time, communication with friends on readings), and the degree to which students like reading. Furthermore, the heterogeneous analysis suggests that disadvantaged (male, low-income, low-performing at baseline) students benefited more in terms of reading time, communication with friends on reading, and affinity towards reading. However, we did not find any overall effect of the program on other variables such as student confidence in reading, reading achievement, creativity, or performance in math or Chinese class. The heterogeneous analysis shows that while students, in general, did not benefit, male students did benefit in terms of confidence in reading and performance in math. A follow-up evaluation of the program is suggested to examine whether the program improves student reading achievement or not in the longer run. Acknowledgement : This work was supported by the Xinhe Foundation. We are grateful for the encouragement and support from Xin Liu and Song Wu. Shoulder Action has provided invaluable support for our field work. The authors are supported by the 111 Project (Grant number B16031).