METHODS The tuberculin skin test (TST) is often included in diagnostic algorithms for tuberculosis (TB) in children. TST interpretation, however, may be complicated by prior Bacillus Calmette-Guerin (BCG) vaccination. We assessed the prevalence of and risk factors for positive TST reactions in children 3 to 60 months of age in Botswana, a country with high TB rates and BCG coverage of over 90%. METHODS A multi-stage cluster survey was conducted in one rural and three urban districts. Data collected included demographic characteristics, nutritional indices, vaccination status, and prior TB exposure. Mantoux TSTs were administered and induration measured at 48-72 hours. RESULTS Of 821 children identified, 783 had TSTs placed and read. Of the 759 children with vaccination cards, 755 (99.5%) had received BCG vaccine. Seventy-nine per cent of children had 0 mm induration, 7% had > or =10 mm induration ('positive' TST), and 2% had > or =15 mm. A positive TST was associated with reported contact with any person with active TB (odds ratio [OR] 1.9; 95% confidence interval [CI] 1.02-3.6), or a mother (OR 5.1; 95% CI 2.1-12.4) or aunt (OR 5.3; 95% CI 2.0-14.0) with active TB. TSTs > or =5 mm (but not > or =10 mm) were associated with presence of a BCG scar. Positive reactions were not associated with age, time since BCG vaccination, clinical signs or symptoms of TB, nutritional status, crowding, or recent measles or polio immunization. CONCLUSIONS The TST remains useful in identifying children with tuberculous infection in this setting of high TB prevalence and extensive BCG coverage.