Community characteristics have been associated with racial and ethnic health disparities for a wide range of ailments and conditions. Previous research has found that rates of AIDS cases among injection drug users (IDUs) vary by community characteristics. However, few studies have examined whether community characteristics are associated with HIV risk behaviors among IDUs. To address this gap in the literature, we examined the associations between census-tract-level community characteristics and injection-related and sex-related HIV risk behaviors among IDUs in the San Francisco Bay Area. Individual HIV risk behaviors were collected from 4,956 IDUs between 1998 and 2002. Using 2000 US census data, we constructed four census-level community measures: percent African American, percent male unemployment, percent of households that receive public assistance, and median household income. All community variables were measured continuously. Multilevel modeling was used to determine if community characteristics were associated with recent (in the last 6 months) receptive and distributive syringe sharing, multiple sex partners, and unprotected sex risk while controlling for potential individual-level confounders. In bivariate analysis, most of the census-tract-level community characteristics were significantly associated with injection-related HIV risk, while no community characteristics were associated with sex-related risk. However, results from multivariate multilevel models indicate that only percent African American in a census tract was associated with receptive [adjusted odds ratio (AOR) = 0.93; 95% confidence interval (CI) = 0.89, 0.99] and distributive syringe sharing (AOR = 0.94; 95% CI = 0.92, 0.99), net of individual-level characteristics. Accounting for individual-level factors in the multivariate model in the sex-related risk models revealed a significant inverse relationship between percent African American and propensity to engage in unprotected sex (AOR = 0.95; 95% CI = 0.92, 0.99); community-level characteristics remained unassociated with multiple sex partners. In this exploratory analysis, percent African American in a census tract was inversely associated with injection-related risk. The census-tract-level community characteristics we examined seem to exert little influence on individual risk among long-term chronic IDUs. More research is needed examining the influence of other community characteristics that were unmeasured in this paper but might be related to sex and drug risk among IDUs such as shooting galleries, crack houses, drug markets, and availability of preventive HIV services.