This paper reviews the demographic evidence for the relationship between male circumcision and HIV infection in national or sub-national African populations. A meta-analysis based on 18 countries, representing more than half of the population of sub-Saharan Africa, shows no relationship [standardized odds ratio?=?1.00; 95% CI: 0.96–1.05]. There were even more countries in which HIV prevalence was higher among circumcised persons than countries where it was lower. In only five countries, the odds ratio of HIV prevalence (circumcised/intact) was significantly different from 1.0; three countries where it was higher, and two countries where it was lower. The contrast between lack of demographic impact and results from clinical trial is striking, and can probably be explained by the low clinical efficacy in situations of intense and repeated exposure, and by the interactions with the many other determinants of HIV spread. This paper also addresses some ethical and political issues, and in particular raises the question of power abuse, which may lie in the practice of genital mutilations and relevant international recommendations.