There is limited research on within-person clustering of early-life behavioral issues and risk for alcohol use issues. This study examined the relevance of childhood behavior problem clusters for predicting alcohol problems in emerging adulthood. This study examined full behavioral profiles as predictors of alcohol-related problems, rather than individual behaviors. The Pathways to Desistance dataset comprised of 1354 justice-involved youth followed across seven years post-adjudication collected from Maricopa County, Arizona and Philadelphia Pennsylvania were analyzed. Latent class analysis was used to identify clusters of childhood behavioral problems (cheating, disturbing class, getting drunk/stoned, stealing, fighting). Regression modeling was used to examine various alcohol use outcomes. Findings indicated that a three-class model best fit the data (Low; FDC; High). Assignment to the High class predicted greater social consequences of alcohol use, but this effect was only marginally significant. No significant effects were observed for either alcohol dependence measure.