Aims. To examine the injecting and sexual risk behaviours of gay men in London who use anabolic steroids or other fitness-enhancing substances (referred to as AS). Design. Cross-sectional survey using self-administered questionnaire. Setting. Five gyms in central London. Participants. 1004 gay men using the gyms during September-October 1997. Measurements. Proportion of men who report (i) injecting AS, (ii) sharing needles and (iii) HIV status-unknown unprotected anal intercourse (UAI). Findings. Of 1004 men, 136 (13.5%) were current users of AS (range across the five gyms, 2.7-21.2%, p < 0.001), and 81 (8.1%) injected AS (range 0.4-17.5%, p < 0.001). None said they shared a needle with other users and more than 90% said they always used disposable units from sealed packets. Among current AS users, 20.9% (28/136) reported status-unknown UAI compared with 12.9% (107/827) of never-users (p = 0.02), a differential which remained significant after adjusting for confounding factors (adjusted odds ratio = 1.75, 95% CI 1.05, 2.91 , p = 0.03). Conclusions. While nearly one in 10 gay men in this study injected anabolic steroids or other fitness-enhancing substances, none reported sharing needles. Steroid users were, however, more likely to report status-unknown UAI than other men, a differential that remained after adjusting for confounding factors. Since any change in injecting practice could dramatically increase the risk of HIV transmission in this population, behavioural surveillance to monitor risk behaviours among gay men using anabolic steroids is recommended, as are targeted HIV prevention programmes.