Studies in heterosexual HIV serodiscordant couples have provided critical evidence on the role of HIV treatments in reducing HIV transmission risk. However, there are limited data regarding the effect of treatment on HIV transmission in homosexual male couples. We examined features of male homosexual HIV serodiscordant relationships that may impact upon the design of HIV treatment and transmission studies.Data were from a prospective cohort study of HIV-negative homosexual men in Sydney, Australia. Men were followed up with six-monthly interviews and annual testing for HIV. Characteristics of men in HIV serodiscordant and seroconcordant relationships at baseline were compared, and a longitudinal analysis performed of rate of relationship break-up and of HIV incidence.At baseline, 5.5% of participants (n=79) had an HIV-positive partner. Most (80.8%) of these relationships were non-monogamous, and 36.7% of men reported recent unprotected anal intercourse (UAI) with casual partners. The rate of relationship break-up was 29.5 per 100 person-years. Half of men in serodiscordant relationships (49.4%) reported recent UAI with their regular partners. HIV incidence was 2.2 per 100 person-years. It was substantially higher in relationships of less than one year's duration (6.1 per 100 person-years) and in men who reported unprotected receptive anal intercourse with ejaculation with their regular partners (15.5 per 100 person-years).Levels of HIV transmission risk and incidence were high, particularly in early relationships. Rates of relationship break-up were high. These data suggest that studies of HIV treatments and transmission in homosexual serodiscordant couples should focus on early relationships so as not to underestimate risk, and sample sizes must allow for high rates of relationship break-up.