Few types of blood exposures have been assessed in relation to incident HIV infection in sub-Saharan Africa, despite evidence that penile-vaginal sex cannot account for the epidemic in the region. To investigate correlates of incident HIV infection in Calabar, Nigeria, we surveyed clients at voluntary HIV counselling and testing centres. Participating clients who tested multiple times were generally similar to those testing only once in terms of demographic characteristics, sexual and blood exposures and HIV prevalence. Blood exposures were common. Serial testers had a 10% annual incidence of HIV infection. Seroconverters and seronegative serial testers were similar on most demographic characteristics and sexual exposures. However, seroconverters were more likely than seronegatives to report blood exposures during the test interval, both for most specific exposures as well as summary measures of blood exposures. In particular, seroconverters were substantially more likely to report one of a set of blood exposures that cannot be explained as a consequence of unprotected vaginal sex or of health care for symptoms of HIV infection (adjusted odds ratio = 6.6, 95% confidence interval = 1.2-38). The study design we used is an inexpensive approach for describing the local epidemiology of HIV transmission and can also serve as the foundation for more definitive investigations that employ contact tracing and sequencing of HIV DNA.