Over the past decade, a number of epidemiological studies have found a direct negative correlation between the practice of male circumcision (MC) and the prevalence of HIV in the general population. These studies were supported by three Randomised Control Trials in Kenya, Uganda and South Africa that found that MC can reduce the probability of female to male infection by between 51 and 60%. Given this, male circumcision is becoming an increasingly discussed addition to HIV prevention programmes in sub-Saharan Africa. Using DHS data from Swaziland and Lesotho and BAIS II data from Botswana this paper uses multivariate logistic regressions to assess the impact of MC on HIV. It further tests for any sample selection bias within the data as a result of non-response of HIV testing. The efficacy of MC in reducing the risk of HIV transmissions is dependent on circumcised men continuing to practice safe sex such as a single partner and condom use. To test for sexual risk a further logistic regression is run using condom use as the dependent variable to determine whether circumcised men engage in riskier sexual behaviour. After controlling for demographic, socio-economic and behavioural determinants of HIV, MC is found to be significant in reducing the probability of being HIV positive in Botswana and Swaziland. In these two countries circumcised men do not appear to engage in riskier sexual behaviour. In Lesotho, despite the high prevalence of MC it does not appear significant in explaining HIV status. This may be explained in part by the fact that circumcised men have a lower probability of using condoms, increasing their risk and negating the effect of MC. Other factors which may explain this relationship in Lesotho is the possibility that traditional circumcision does not remove the entire foreskin and the age at which the circumcision is taking place. All these need to be considered if implementing a large scale MC programme.