ObjectiveTo assess the importance of community-recruited drug injectors with no treatment experience and no previous testing history when estimating HIV prevalence among drug injectors.DesignHIV testing behaviour and prevalence were measured in a serial point prevalence study of drug injectors recruited in community-based non-treatment and treatment settings.MethodsPearson's χ2and Fisher's exact tests were used to measure differences between treatment groups and non-treatment groups in demographic characteristics, including age and length of injecting career, recent drug use, recent syringe sharing, uptake of previous HIV testing, confirmed HIV-antibody status by saliva and self-reported HIV status.ResultsRates of HIV-antibody testing were significantly lower in community-recruited drug injectors with no experience of treatment than those previously or currently in treatment. Confirmed HIV prevalence by saliva was highest in drug injectors with no experience of treatment. HIV-antibody-positive drug injectors with no treatment experience were less likely to be aware of their antibody status than injectors who were, or who had been, in treatment.ConclusionsPrevious studies of HIV prevalence among drug injectors may be biased by drawing on samples primarily from treatment settings. Drug injectors with no treatment experience and no previous history of HIV testing should be included in HIV prevalence studies. Regular and repeat HIV testing in low-threshold community-based programmes should be considered a necessary part of interventions which seek to provide better access to treatment and other HIV prevention services.