This is one of the first studies to evaluate the impact of personalized learning delivered through technology in a developing country. We randomly assigned 1,528 students in grades 6-8 in 15 “model” public schools who were using a computer-adaptive learning software to: (a) a control group, in which they were only able to access the activities for their enrolled grade level (i.e., the norm for most software products evaluated in developing countries); or (b) a treatment group, in which they were assigned exercises appropriate for their individual preparation level, across a wide range of grade levels, based on a diagnostic test. After nine months, personalized learning had a null effect on the math achievement of the average student. However, treatment students with low initial performance outperformed their control counterparts by 0.22 standard deviations. Our results suggest that personalized learning is most beneficial for relatively low-performing students, who need help to catch up with their peers.