OBJECTIVES: To examine the determinants of uptake of voluntary counselling and testing (VCT) services, to assess changes in sexual risk behaviour following VCT, and to compare HIV incidence amongst testers and non-testers. METHODS: Prospective population-based cohort study of adult men and women in the Manicaland province of eastern Zimbabwe. Demographic, socioeconomic, sexual behaviour and VCT utilization data were collected at baseline (1998-2000) and follow-up (3 years later). HIV status was determined by HIV-1 antibody detection. In addition to services provided by the government and non-governmental organizations, a mobile VCT clinic was available at study sites. RESULTS: Lifetime uptake of VCT increased from under 6% to 11% at follow-up. Age, increasing education and knowledge of HIV were associated with VCT uptake. Women who took a test were more likely to be HIV positive and to have greater HIV knowledge and fewer total lifetime partners. After controlling for demographic characteristics, sexual behaviour was not independently associated with VCT uptake. Women who tested positive reported increased consistent condom use in their regular partnerships. However, individuals who tested negative were more likely to adopt more risky behaviours in terms of numbers of partnerships in the last month, the last year and in concurrent partnerships. HIV incidence during follow-up did not differ between testers and non-testers. CONCLUSION: Motivation for VCT uptake was driven by knowledge and education rather than sexual risk. Increased sexual risk following receipt of a negative result may be a serious unintended consequence of VCT. It should be minimized with appropriate pre- and post-test counselling.