An 8-year research-service collaboration in tuberculosis (TB) in Hlabisa has yielded important data on the emerging HIV-related TB epidemic in South Africa, demonstrating a substantial increase in caseload, HIV prevalence and mortality. A successful, cost-effective, community-based programme of directly observed therapy, utilising volunteers, clinics and community health workers, has helped ensure adherence to therapy. As the caseload has increased, however, case-holding has started to fall, suggesting the need for more effective defaulter-tracing strategies. More fundamental work has studied the molecular epidemiology of TB, suggesting that much of the disease is recently, and casually, transmitted. This collaborative experience may be a model for efforts to control other important public health problems.