Abstract Background This study examines the trajectory of social withdrawal from the fourth grade of elementary school to the first grade of middle school in Korea. By classifying them, we identify individual differences such as individual, family, and school factors. Methods This study uses data from the Korean Youth Panel Survey (KYCPS) conducted by the Korea Youth Policy Research Institute, which includes data from the 1st-4th waves of KYCPS 2018, with a total of 2,607 participants (1,313 boys and 1,294 girls). A latent class growth analysis verifies the trajectory of youth social withdrawal, classifies latent classes, and identifies predictive. SPSS 24.0 was used to analyze the main variables’ characteristics, descriptive analysis, and reliability. Mplus 8.9 was used to test the research model. Results First, results indicated that three was the optimum number of latent classes. We identified three latent classes ‘high level of withdrawal continued (HL)’, ‘medium level of withdrawal increased (ML)’, and ‘low level of withdrawal increased (LL)’. Second, when compared the ‘LL’ and ‘HL’ groups, the higher the self-esteem, the higher the friendship, and the higher the dependence on smartphones, the more likely it was to be in the ‘HL’ group. Third, when comparing ‘LL’ and ‘ML’ groups, the more negative the parenting attitude, the more likely it was to be in the ‘ML’ group. On the other hand, higher self-esteem and more positive friend and teacher relationships were associated with a higher likelihood of being in the ‘LL’ group. Conclusions It is necessary to seek differentiated interventions to prevent social withdrawal among early adolescents. Parental education to understand the adolescent period and practical approaches to intervene in the negative psycho-emotional aspects of adolescents continuously are needed to enhance the healthy psychological development of early adolescents. Key messages • The higher the self-esteem, the higher the friendship, and the higher the dependence on smartphones, the more likely it was to be in the ‘high level of withdrawal continued’ group. • Higher self-esteem and more positive friend and teacher relationships were associated with a higher likelihood of being in the ‘low level of withdrawal increased’ group.