Purpose: This study was to identify the effects of the predictive factors of Internet addiction and to develop a predictive model that explains Internet addiction among adolescents in Korea. Method: Data was collected from 664 adolescents in Puasan and Ulsan. Data analysis was done using the SPSS 10.0 Win Program for descriptive and correlational statistics and the LISREL 8.53 Win Program for Covariance structural analysis. Result: The fit of the hypothetical model to the data was moderate, it was modified by deleting four paths. Compared to the hypothetical model, the revised one had a better fit to the data($chi^2$/df=6.50, GFI=.99, AGFI=.90, RMR=.02, NNFI=.88, NFI=.98, CN=279). Gender, friend support, and loneliness had significant direct effects on Internet addiction. The direct and indirect effect of self-esteem on Internet addiction was significant. The indirect effect of family support on Internet addiction was significant but its direct effect was insignificant. The direct and indirect effects of stress and impulsiveness on Internet addiction were significant but their total effects were insignificant. Conclusion: Finally, to decrease loneliness, and to increase family support and self-esteem, an effective intervention program and education should be developed to prevent Internet addiction of adolescents.