Education of poor and disadvantaged populations has been a long-standing challenge for education systems in both developed and developing countries. In China, millions of students in rural areas and migrant communities lag far behind their urban counterparts in terms of academic achievement. When they fall behind, they often have no way to catch up. Many of their parents have neither the skills nor the money to provide remedial tutoring; rural teachers often do not have time to give students the individual attention they need. Given this, there is growing interest by both educators and policymakers in helping underperforming students catch up using computer assisted learning (CAL). While CAL interventions have been shown to be effective internationally and elsewhere in China, traditional software-based CAL programs are difficult and costly to implement. An online version of CAL (OCAL), however, may be able to bypass many of offline CAL’s implementation problems and enhance the remedial tutoring experience. Unfortunately, there is little empirical evidence on whether OCAL programs can be effective in improving the quality of rural primary school education in developing countries. The objective of this paper is to examine the impact of an OCAL intervention on the academic and non-academic performance of students and to explore the mechanism behind OCAL’s impact. Importantly, we also aim to assess the cost effectiveness of the new OCAL program versus traditional CAL interventions. To achieve these objectives, we carried out a randomized controlled trial (RCT) involving over 1650 fifth grade students in 44 schools in rural areas and migrant communities across China. Students in the 22 treatment schools attended two 40- minute OCAL sessions during their computer class each week for one semester; the students in the other 22 schools were in the control group and did not receive any intervention. According to our findings, OCAL improved overall English scores of students in the treatment group relative to the control group by 0.56 standard deviations. This impact is large when compared with offline CAL programs. We found that OCAL also led to a positive change in the attitudes of students towards English learning and towards student aspirations for their future education level. We found three possible explanations for OCAL’s impact. After rejecting the possibility of the Hawthorne Effect or self-efficacy-induced changes, we believe interest-oriented stimulation is the main source of improvement among students. The chance for comparison and competition with peers, as well as customized remedial question banks tailored to each student’s individual needs, likely contributed to the measured increases in academic performance among students in our sample. Cost-effectiveness analysis showed that the OCAL program is more cost-effective than traditional offline CAL, a comparison which is significant for policymakers as it indicates high potential for OCAL program expansion.