In the last decade studies have identified a myriad of factors associated with continued risky sex in gay men and men who have sex with men. More recently the phenomenon of 'relapse' has been identified among this population. The results of such studies have been fed into the ongoing development of strategies to prevent HIV transmission. This paper critically examines this research--most of which is quantitative--and argues that the usefulness of the research is limited. Because of methodological problems, the evidence is contradictory and indeed, comparison across studies is almost impossible. An examination of the concept of 'relapse' illustrates the problems of relying on quantitative research which, in the main, removes individuals from their social context. The paper concludes that more qualitative research is required to locate individuals within their social milieu and to better understand them as individuals interacting with others and involved in an ongoing decision-making process about sexual pleasure and risk.