OBJECTIVES: To assess factors associated with youth substance abuse (SA) using a U.S. representative sample. METHODS: A cross-sectional study using the 2012 US National Survey on Drug Use and Health (NSDUH) database was performed on youth population 12 to 17 years old (n=14,600).  A structural equation model was applied to investigate factors that influenced the youth SA, the dependent latent variable, described using indicators including tobacco, alcohol and marijuana use. Six independent latent factors were youth tolerant attitude towards SA, perceived risk of SA, pressure from social norm, risk-prone behavior, adverse school environment of SA, and family support (e.g. parental care and concern).   RESULTS: The model fitness showed the internal structure of the model was acceptable (RMSR=0.064; CFI=0.937). The observed indicators were sufficiently explained by its respective latent variables (e.g. 67% to 72% of variances for indicators of school environment were explained). All independent latent variables were statistically significantly associated with SA (P<0.01). The perceived risk of SA, pressure from social norm and family support were negatively associated with SA. On the other hand, youth with tolerant attitude toward SA, risk-prone behavior and adverse school environment were positively associated with SA.  CONCLUSIONS: Findings suggested that effective prevention of youth SA can be made from increasing the awareness of the risk of SA, improving school environment and enhancing family support. In particular, we should target the youth population with risk-prone behavior and tolerant attitude towards SA.