Following latent class analysis (LCA) approach we examined patterns of HIV risk using two related domains of behavior: drug use, and sexual activity among 523 injection drug users (IDUs) recruited into the 2009 National HIV behavioral surveillance system. Using posterior probability of endorsing six drug and sexual items, we identified three distinct classes representing underlying HIV risk. Forty percent of our participants were at highest risk, 25 % at medium risk, and 35 % at lowest risk for HIV infection. Compared to the Lowest-risk class members, the Highest-risk class members had riskier drug and sexual behaviors and had higher prevalence of HIV cases (6 vs. 4 %). This analysis underscores the merit of LCA to empirically identify risk patterns using multiple indicators and our results show HIV risk varies among IDUs as their drug and sexual behaviors. Tailored and targeted prevention and treatment interventions for the dual risk pattern are required rather than for drug or sexual risk in silos.