The objective of our study was to estimate the prevalence of HIV infection among pregnant women in Bobo-Dioulasso (Burkina Faso) according to 2 survey methods. Unlinked anonymous HIV screening was performed among women attending 2 antenatal clinics. Voluntary and confidential HIV counselling and testing were offered to women attending 2 other antenatal clinics in the same time period, September-October 1996. Voluntary HIV testing was performed in the context of a clinical trial on mother-to-child transmission of HIV (ANRS 049 clinical trial) with an acceptance rate of HIV testing of 93%. The first survey recruited 200 women and the second, 424. The mean age (24.6 years vs 24.8 years) and the mean number of pregnancies (3.1 vs 3.3) of women were comparable, in the 2 studies (P=0.69 and P=0.26, respectively). Prevalence of HIV infection in the blinded survey was estimated at 10.0% (95% confidence interval (CI): 6.4-15.2), while it was 9.4% (95% CI: 6.9-12.7) in the voluntary HIV screening programme. These 2 estimates were not statistically different (P=0.82). In the voluntary screening study, the prevalence of HIV infection was significantly different between age groups 15-24 years and 25-49 years (13.9% vs 4.5%, P < 0.001). In the age group 25-49 years, the prevalence of HIV infection estimated in the blinded study and in the voluntary screening study were significantly different (10.5% vs 4.5%, P=0.04) suggesting a potential participation bias among pregnant women of older age in the voluntary, confidential HIV screening group. In conclusion, for the purpose of HIV surveillance, the most reliable method for HIV prevalence remains the unlinked, anonymous testing.