Past research on adolescents’ addiction to Internet had revealed that most of those addicts were low in self-concept and showed withdrawal behavior in social interactions. And among college students who were addicts, there was reported connection between online experience and addictive behavior. To examine effects of self-concept and online experience, to be more specific, the negative online experience, on addictive behaviors while surfing Internet, we conducted the present research in adolescents, who emerged as the predominant group found to be Internet addicts in modern China. Besides, we also investigated the possible influence of school environment given the hot issue on “key school policy” proposed right after the establishment of new China in 1950s. Participants included 4,094 adolescents (1864 males, 2078 females), of which 1830 came from primary school from grade four to grade six and 2263 students were from middle school. We conducted a large online survey on adolescents’ Internet addictive behaviors across 10 different domestic regions, for instance Guizhou, Chongqing, and Jiangsu, etc. The instruments used in this study included the Differentiation of School and Society Situation for Chinese primary and middle school students, which assessed adolescents’ self-concept, the Negative Online Experience scale and Internet Addiction Diagnostic questionnaire, which measured degree of addiction to Internet. The results from current study revealed that adolescents’ self-concept was significantly negatively related to their degree of Internet addiction (β=-.137, p<.001), and their negative online experience could significantly positively predict their degree of Internet addiction (β=.281, p<.001). With regard to school type, there was also a significant main effect (β=.051, p=.001), indicating that students from key schools were more vulnerable to Internet addiction than those from non-key schools. Moreover, there was a significant three-way interaction among self-concept, negative online experience and school type, B=-.064, t (1, 3918) =-2.218, p=.027, η2=.001, such that for students from key schools, their addiction to Internet could be negatively predicted by levels of self-concept but in non-key schools the moderating effect of negative online experience should be carefully examined. The findings had two implications on the issue of adolescents’ addiction to Internet. First, it suggested that adolescents’ online experience could have magnificent impact on their subsequent addicted behavior to Internet, and more importance should be attached to adolescents’ way of handling online interactions and thereby provided necessary and efficient guidance. Second, there was an innovative finding concerning the school environment in which adolescents inhabited. The results demonstrated that students from key schools reported more addictive behaviors than their counterparts, which meant that there was probably some lacking in social activity due to extra burden in key school. For intervention, we should focus on different aspects in terms of different school types. Specifically, for key schools, there should be alternative entertainment; and for non-key schools, there should be educational programs on behavioral regulation while surfing Internet. Despite several limitations, especially with regard to the failure of establishing casual relations in this correlational study, the present work contributed, to a certain extent, to the interpretation and potential intervention of adolescents’ Internet addiction, as well as some insight in the improvement of adolescents’ school environment.