OBJECTIVES: To explore the role of male circumcision in the spread of HIV infection in four urban populations in sub-Saharan Africa. DESIGN AND METHODS: A cross-sectional population based study was conducted in four cities in sub-Saharan Africa with different levels of HIV infection. HIV prevalence among adults was relatively low in Cotonou (Benin) and in Yaoundé (Cameroon), and exceeded 25% in Kisumu (Kenya) and in Ndola (Zambia). In each city, a random sample was taken of men and women aged 15-49 years from the general population. Consenting study participants were interviewed about their sociodemographic characteristics and their sexual behaviour, and were tested for HIV, herpes simplex virus type 2, syphilis, gonorrhoea and chlamydial infection. Men underwent a genital examination. RESULTS: In Cotonou and in Yaoundé, the two low HIV prevalence cities, 99% of men were circumcised. In Kisumu 27.5% of men were circumcised, and in Ndola this proportion was 9%. In Kisumu, the prevalence of HIV infection was 9.9% among circumcised men and 26.6% among uncircumcised men. After controlling for socio-demographic characteristics, sexual behaviour and other sexually transmitted infections, the protective effect of male circumcision remained with an adjusted odds ratio of 0.26 (95% confidence interval = 0.12-0.56). In Ndola, the prevalence of HIV infection was 25.0% in circumcised men and 26.0% in uncircumcised men. The power was insufficient to adjust for any differences in sexual behaviour. CONCLUSIONS: The differences in epidemic spread of HIV are likely to be due to differences in the probability of transmission of HIV during sexual exposure as well as differences in sexual behaviour. Male circumcision is one of the factors influencing the transmission of HIV during sexual intercourse, and this study confirms the population level association between HIV and lack of male circumcision, as well as a strong individual level association in Kisumu, the only city with sufficient power to analyze this association.