Objectives: Most data on HIV prevalence in low-risk populations in sub-Saharan Africa are drawn from sentinel surveys of pregnant women attending antenatal clinics and are not representative of formal sector workforces. We surveyed workforces in southern Africa to determine HIV prevalence among formally employed, largely male populations. Methods: Voluntary, anonymous, unlinked seroprevalence surveys of 34 workforces with 44,000 employees were carried out in South Africa, Botswana, and Zambia in 2000-2001. Results were stratified to obtain estimates of prevalence by industrial sector, location, age, sex, and job level. Results: Average HIV prevalence for the entire sample was 16.6% (95% CI: 16.3-17.0%). Country-wide prevalence was 14.5% (14.1-14.9%) in South Africa, 17.9% (17.1-18.7%) in Zambia, and 24.6% (23.6-25.7%) in Botswana. Among industrial sectors, mining (18.0%, 17.6-18.5%) and metal processing (17.3%, 15.9-18.7%) had the highest infection rates. Males, who comprised 85% of participants of known sex, were more likely (16.3%, 15.3-17.4%) to be infected than were females (10.7%, 8.7-12.7%). Contract (23%, 21.9-24.1%), unskilled (18.3%, 17.5-19.1%), and semi-skilled workers (18.7%, 18.1-19.4%) were much more likely to be infected than were skilled workers (10.5%, 9.5-11.4%) and managers (4.5%, 3.4-5.6%). Participation in the surveys averaged 63% of eligible employees. Conclusions: HIV prevalence among formally employed workers in southern Africa shows different patterns than among antenatal clinic attendees. Anonymous workplace surveys generate prevalence estimates for demographic groups that are not represented in antenatal surveys and can strengthen support for prevention and treatment interventions.