Objective To assess the prevalence and characteristics of unintentional injuries among 5 to 9 grades students for both urban and migrant children groups, and probe the major socio-demographic and psychosocial risk factors. Methods A cross-sectional survey was conducted from May to June in 2013 with 2266 urban and 1770 migrant children among 5 to 9 grades from 13 schools in Yinzhou, Southeast China. Unintentional injuries and psychosocial factors were measured via self-administered questionnaires. Descriptive statistics and binary logistic regression were employed in data analysis. Results The annual prevalence of unintentional injury in the past year was 38%. Compared with urban children, migrant children reported lower prevalence of sprains/fractures (15.5 vs. 18.7%), but higher in animal bites (15.7 vs. 9.6%), burns (20.7 vs. 16%), nonfatal drowning (6 vs. 2.7%), and unintentional injury (40.9 vs. 35.7%). Binary logistic regression analysis revealed that male, young age, attending migrant children school, scoring high in emotional symptoms, conduct problems or hyperactivity, with suicide ideation, and experiencing verbal maltreatment were associated with unintentional injury incidence in the past 12 months. Conclusions The annual prevalence of unintentional injury among 5 to 9 grades schoolchildren in Yinzhou was high. Child unintentional injuries were associated with multi-level psychosocial factors. Although the direct correlation between migrant status and unintentional injuries was unobserved, migration could have an indirect relationship with injury. The findings could inform some implications for prevention strategies in urban area. Keywords Unintentional injury • Mental health • Child maltreatment • Migrant children • China Unintentional injury has been recognized as a pervasive challenge in public health, raising great concern globally (Chandran et al. 2010; de Ramirez et al. 2012). A previous study estimated a nationwide injury death rate of 65 per 100000 population in China (Wang et al. 2008), which was