OBJECTIVE: Tuberculosis (TB) infection is highly prevalent in developing countries. As infected children represent a large proportion of the pool from which TB cases will arise, knowledge of the factors that influence TB infection in children are of importance to evaluate transmission of infection in the community and adapt TB control activities. There are limited data on the risk of infection in child populations in developing countries. METHODS: We performed a household contact study in The Gambia (West Africa), in which children who were living in contact with individuals who had proven smear-positive pulmonary TB cases were investigated. A questionnaire was addressed to the mother or caregiver of the child to investigate the presence of various risk factors and assess the degree of exposure of the child to the individual with TB within the household. A tuberculin skin test (TST) was performed on each child. TST sizes > or =5 and 10 mm, respectively, were considered positive. RESULTS: Households of 206 TB cases were visited, and 384 children aged <5 years were examined. The median age was 2, and 48% were girls. The distribution of TST responses followed a bimodal pattern, with 135 (35%) children presenting a palpable induration. Random effects logistic regression analysis demonstrated that the risk of positive TST response in the child increased with the geographic proximity of the child to the individual with TB within the household and with the degree of activities shared with the individual with TB. It was also associated with the clinical severity of the disease in the index case. Nutritional status and presence of a bacille Calmette-Guérin (BCG) scar were not independent risk factors for TST positivity in this population. On multivariate analysis, the effect of geographic proximity to the individual with TB, household size, and duration of cough in the index case persisted for TST responses > or =5 mm. CONCLUSIONS: In a highly endemic country with high BCG vaccination coverage in Africa, TB infection in children who were in contact with individual with infectious TB was directly related to the intensity of exposure of the child to the individual with TB. Our data suggest that a positive TST in a child reflects most probably TB infection rather than previous BCG vaccination. Contact tracing can play a major role in the control of TB in developing countries.