We know that peer education, or the use of popular opinion leaders (POLs), works in terms of reducing reported risk behaviour for HIV infection amongst gay men. The work of Jeffrey Kelly and his colleagues provides some of the best scientific evidence in support of this approach. Influenced by this work, we undertook a peer education intervention amongst gay men in bars in Glasgow--the Gay Men's Task Force (GMTF)--but failed to demonstrate any reduction in sexual risk behaviour for HIV infection. In this paper we describe why we were unable to repeat in Scotland the success in small cities in the USA of the POL model. Our explanations include: failure to replicate the 'core elements' of POL; spatial and temporal differences between the original POL settings and the bars of Glasgow; and the currency of ideas such as 'peer education' beyond the protocols designed for their implementation. However, we also describe some of the successful features of the GMTF in Glasgow, and the continued value of peer education in contributing to reductions in sexual risk behaviour for HIV infection.