methodology to assess both HIV prevalence and risk factors.

: This is the first study of men who have sex with men in Africa to use respondent driven sampling (RDS) methodology to assess both HIV prevalence and risk factors. Intriguingly, the multivariate analysis revealed that the strongest adjusted odds ratio for associations with decreased risk of HIV infection was for being circumcised. Insertive anal intercourse was reported more commonly than receptive anal intercourse (85.2% ever versus 20.6% ever) and HIV prevalence among gay-identified men was more than 3 times that of bisexual and straight-identified men who have sex with men. Since the latter groups have to date been more likely to report the exclusive practice of insertive anal intercourse with male partners, this study may be providing the hypothesis-generating evidence that would support a trial of male circumcision for HIV prevention among primarily insertive men who have sex with men. On the more immediately practical side, rapid research is needed to find out whether current messages about male circumcision for HIV prevention are reaching bisexual and straight-identified men in South Africa and other high HIV prevalence settings.