The purpose of this study was to classify latent group according to changes of internet usage pattern based on the degree of internet usage. At the same time, the study sought to validate differences of variables such as level of individual and school environment between latent groups classified by longitudinal internet using patterns. This study used longitudinal data(2010-2013) "Korean Children and Youth Panel Study(KCYPS) conducted by National Youth Policy Institute targeting middle school students. In examing hypotheses, Latent Class Analysis(LCA) and Analysis of Variance (ANOVA) were employed. The results are as follows. First, it was divided in to 4 latent groups based on longitudinal internet using patterns and each of them were designated as "High-Usage Group", "Low-Usage Group", "Informative-Usage Group", and "Declining Group". Second, differences of individual factors (depression, social withdrawal, aggression, somatic symptoms) and school environment factors (learning activity, regulations, peer relationships, student-teacher relationships) between latent classes based on longitudinal internet using patterns were statistically significant. Implications of the study and future for future research are discussed.