BACKGROUND:Previous mathematical models have indicated that any decrease in HIV incidence in homosexual men due to decreased infectiousness from antiretroviral treatment (ARV) may be offset by modest increases in unsafe sex. The aims of this study were to assess the effects of ARV use and increasing unprotected anal intercourse with casual partners (UAIC) in homosexual men on HIV incidence during 1995-2001 and to project HIV incidence depending on trends in ARV use and UAIC. METHODS:A mathematical model of HIV transmission among homosexual men in Australia was developed. HIV incidence during 1995-2001 was estimated assuming that 70% of men in whom HIV was diagnosed received ARVs and assuming a 10% annual increase in UAIC. For 2001-2006, scenarios included ARV levels remaining at 70% or declining to 50% by 2006, combined with UAIC levels remaining at the 2001 level or continuing to increase annually by 10%. FINDINGS:The number of incident HIV cases per year was predicted to have declined during 1996-1998 due to the introduction of effective ARVs, with a slow increase during 1998-2001 due to increased levels of UAIC when use of therapies was fairly stable. From 2001, a continued increase in UAIC was predicted to lead to a rise in HIV incidence. A rise in UAIC combined with a moderate decline in ARV use could lead to a 50% increase in HIV incidence by 2006. INTERPRETATION:These models suggest that widespread ARV use has had some effect in reducing HIV incidence among homosexual men in Australia. However, if current trends in UAIC and ARV use continue, a resurgent HIV epidemic is predicted.