OBJECTIVE: To assess circumcision status as a risk factor for HIV seroconversion in homosexual men. DESIGN, SETTING AND PARTICIPANTS: The Health in Men (HIM) study was a prospective cohort of homosexual men in Sydney, Australia. HIV-negative men (n = 1426) were recruited primarily from community-based sources between 2001 and 2004 and followed to mid-2007. Participants underwent annual HIV testing, and detailed information on sexual risk behaviour was collected every 6 months. MAIN OUTCOME MEASURE: HIV incidence in circumcised compared with uncircumcised participants, stratified by whether or not men predominantly practised the insertive role in anal intercourse. RESULTS: There were 53 HIV seroconversions during follow-up; an incidence of 0.78 per 100 person-years. On multivariate analysis controlling for behavioural risk factors, being circumcised was associated with a nonsignificant reduction in risk of HIV seroconversion [hazard ratio 0.78, 95% confidence interval (CI) 0.42-1.45, P = 0.424]. Among one-third of study participants who reported a preference for the insertive role in anal intercourse, being circumcised was associated with a significant reduction in HIV incidence after controlling for age and unprotected anal intercourse (UAI) (hazard ratio 0.11, 95% CI 0.03-0.80, P = 0.041). Those who reported a preference for the insertive role overwhelmingly practised insertive rather than receptive UAI. CONCLUSIONS: Overall, circumcision did not significantly reduce the risk of HIV infection in the HIM cohort. However, it was associated with a significant reduction in HIV incidence among those participants who reported a preference for the insertive role in anal intercourse. Circumcision may have a role as an HIV prevention intervention in this subset of homosexual men.