The first HIV-positive diagnosis among injecting drug users (IDU) in Glasgow was made in 1985. By the end of 1987 prevalence among IDU receiving voluntary attributable tests was 4.8%. Since 1990, an annual cross-sectional survey of HIV prevalence and risk behaviours among 500 current Glasgow IDU has provided a more representative sample. Anonymously-tested saliva samples obtained from respondents revealed prevalence rates of 1.8%, 1.2% and 1.0% in 1990, 1991 and 1992 respectively. Since 1987 a wide range of measures aimed at reducing HIV-related risk behaviour among IDU has been introduced and maintained in Glasgow. Against this background, there is evidence that IDU in the city have reduced their risk behaviours. The findings reported here suggest that implementing harm-reduction measures when prevalence is low may inhibit the rapid dissemination of HIV.