Over the last decades, the issue of digital divide has received particular attention from international bodies and researchers in Western countries. One of the main reasons for this growing interest is related to the implications that digital inequalities have for social development and particularly for education. Despite the relevance of the issue, there are still few studies on the digital divide in China and even less on Chinese K-12 schools' students. This paper aims at exploring the digital divide issues among Chinese children from an educational and social perspective. Four schools were selected, involving 658 students aged 10-14. Nine hypotheses on Internet inequality indicators (DiMaggio & Hargittai, 2001) and their possible outcomes were tested. The study found that: (1) students' Internet access at home is better than that at school; (2) compared with parents, teachers have more positive influence on students' Internet behaviour; (3) students from rural or migrant schools score lower on all the Internet inequality indicators (digital access, autonomy of use, social support, Internet use and self-efficacy) and are therefore more disadvantaged in Internet usage status than their urban peers; (4) there are no significant gender differences; (5) the more education parents have received, the better the conditions of their children according to the listed Internet inequality indicators; (6) the dimensions of the Internet inequality indicators (technical apparatus, autonomy of use, availability of social support, variation of use) are significantly correlated to students' Internet self-efficacy; (7) there is high correlation between students' Internet self-efficacy and their exploring behaviours of Internet use and their academic performance. Overall, these results are consistent with data from OECD countries and confirm that the digital divide represents a big social challenge, revealing that schools still have to develop effective strategies to balance social and learning opportunities among students. (Contains 13 tables.)