OBJECTIVE To examine the putative protective effect of HIV-2 infection against subsequent HIV-1 infection. METHODS Retrospective analysis of data from two cross-sectional surveys in the same community. METHODS Two surveys between 1989 and 1998 in a rural area in northwestern Guinea-Bissau provided data from residents aged 15-59 years. HIV testing was done in the first survey. In the second survey, tests were made for both HIV and syphilis, and data on sociodemographic factors and sexual behaviour, including commercial sex work, were gathered. Qualitative polymerase chain reaction amplification of HIV-1 and HIV-2 viral DNA was performed on serologically dually reactive samples. RESULTS Of the 2276 eligible adult villagers initially tested, 60% (1360) provided a second sample. Of 110 HIV-2-infected subjects, 17 became additionally infected with HIV-1 [incidence rate (IR), 26.3/1000 person-years observation]. Of the 1250 HIV-seronegative subjects, 24 became infected with HIV-1 (IR, 2.8/1000 person-years observation). The incidence rate ratio (IRR), comparing the incidence rate in HIV-2-infected people with the rate in HIV-seronegative subjects, was > 1 in all three "risk groups": men, female commercial sex workers, and other women. The overall estimate of the IRR, adjusted for age group and risk group, was 3.24 (confidence interval, 1.5-7.1). CONCLUSIONS There was no protective effect of HIV-2 in this population. HIV-2 cannot be regarded as a vaccine, but, instead, may be a risk factor for HIV-1 infection.