This paper presents the first outcome evaluation of multi-session groupwork for HIV prevention among gay men in the UK. This community-based RCT recruited 50 men, of whom 42% were HIV-positive or untested, and 32% reported status unknown or serodiscordant UAI in the previous 12 months. No knowledge, skills, attitudinal or behavioural differences were detected between intervention and control at baseline. At eight weeks, those attending the group reported significant gains over their control in making sexual choices, physical safety, HIV and STI transmission knowledge, and sexual negotiation skills. At 20 weeks, significant differences remained for HIV and STI transmission knowledge and comfort with sexual choices. Although no behavioural differences were detected, the aims of the National Prevention Strategy were met. This pilot RCT is appraised in the light of modest sample size and attrition, and recommendations for establishing behavioural outcomes are presented. This study has demonstrated that high-risk community samples can be recruited to multi-session interventions, and has provided feasibility data for future rigorous evaluation designs.