Background: Condomless anal intercourse (CLAI) has long been recognised as the primary mode of sexual transmission of HIV in gay and other men who have sex with men (MSM). A variety of measures of CLAI have been commonly used in behavioural surveillance for HIV risk and to forecast trends in HIV infection. However, gay and other MSM’s sexual practices changed as the understanding of disease and treatment options advance. In the present paper, we argue that summary measures such as any CLAI do not accurately measure HIV sexual risk behaviour. Methods: Participants were 1,427 HIV-negative men from the Health in Men cohort study run from 2001 to 2007 in Sydney, Australia, with six-monthly interviews. At each interview, detailed quantitative data on the number of episodes of insertive and receptive CLAI in the last six months were collected, separated by partner type (regular vs. casual) and partners’ HIV status (negative, positive, and HIV status unknown).Results: A total of 228,064 episodes of CLAI were reported during the study period with a mean of 44 episodes per year per participant (median: 14). The great majority of CLAI episodes were with a regular partner (92.6%), most of them with HIV-negative regular partners (84.8%). Participants were more likely to engage in insertive CLAI with casual than with regular partners (66.7% vs. 55.3% of all acts of CLAI with each partner type, p<0.001). Men were more likely to report CLAI in the receptive position with HIV-negative and HIV status unknown partners than with HIV-positive partners (p<0.001 for both regular and casual partners). Conclusion: Gay and other MSM engaging in CLAI demonstrate clear patterns of HIV risk reduction behaviour. As HIV prevention enters the era of antiretroviral-based biomedical approach, using all forms of CLAI indiscriminately as a measure of HIV behavioural risk is not helpful in understanding the current drivers of HIV transmission in the community.