In 2002, we implemented a 4-year HIV prevention intervention for injection drug users (IDUs) in Lang Son Province, Vietnam, and Ning Ming County, Guangxi Province, China, a cross-border region seriously affected by inter-twined epidemics of heroin injection and HIV infection. The interventions involve peer education on HIV risk reduction and provision of new needles/syringes through direct distribution and pharmacy vouchers. We consider this to be a structural intervention in which risk reduction information and sterile injection equipment are diffused through the IDU populations and not limited to those who actually interact with peer educators. The evaluation of structural interventions poses complex methodological challenges. The evaluation of our interventions relies primarily on cross-sectional surveys (interviews and HIV testing) of samples of IDUs selected using a combination of targeted cluster and snowball methods. We consider this to be an appropriate, albeit imperfect, design given the study context. This paper presents analyses of data from the IDU surveys conducted just prior to implementation of the interventions and 24 months thereafter. The cross-border interventions have reached large proportions of the IDUs in the project sites, drug-related HIV risk behaviors have declined in frequency, and HIV prevalence among IDUs has been stable in China and declined in Vietnam over the 24 months since the interventions were implemented. Attribution of these positive trends to the interventions must be qualified in light of possible sampling biases and the absence of control groups. However, we believe that the structural interventions implemented by the cross-border project have played a role in stabilizing HIV prevalence among IDUs two years after they were initiated. Evidence of further diffusion of the interventions among IDUs and continued stability or decline of HIV prevalence would strengthen this case.