Despite medical advances, HIV and AIDS remain incurable. The promotion of safer sexual practices has therefore seemed crucial. However providing information has had limited success in changing behaviour. There are still many new cases of infection each year. In order to help plan effective interventions, researchers have explored peoples attitudes and beliefs around sex, and their associations with behaviour. I suggest that research, through reliance on social cognition models which are individualistic in focus. has neglected wider issues such as socio-cultural contexts of attitudes and behav iour. Such issues are likely to be important. I argue in Chapter One that information about sex and risk has not been delivered ·neutrally·. but rather in the context of messages which contain strong political and social agendas pertaining to sexual it) and moral it). Qualitative work has suggested that issues of pm,er. trust and romance are important in men's constructions and experience of safer sex. Differences have emerged between heterosexual and gay men in these respects. However, because studies have tended to include either gay or heterosexual men. comparisons between heterosexual and gay samples have been difficult. The present qualitative study aimed to explore how men understand HIV risk and safety. It also aimed to explore men's experiences of safer or riskier sexual behav iour. considering relationship and contextual factors as well as individual ones. Semistructured interviews were carried out with heterosexual and gay men in order to allow consideration of any emerging differences between accounts. Interviews were analy sed according to the principles of lnterpretative Phenomenological Analysis. Emerging material was constructed as themes which were eventually organised according to the following Master Themes: Definitions of Risk, Perceived Threat of HIV, Control and Negotiation, What Testing Does and Barriers. These themes are discussed in terms of their implications for intervention. Finally, I discuss quality of the research in terms of issues such as reliability, generalisability and reflectivity.