Setting: Considering the difficulty to establish diagnosis of tuberculosis in childhood, the authors studied three diagnostic criteria published in the literature in the last 20 years, evaluating their sensitivity, specificity, and accuracy. Material and methods: In a prospective study, 94 children at ages ranging from 0 to 15 years having contact with bacillary tuberculous adults were studied in the Municipal Health Center of Duque de Caxias, Rio de Janeiro. The following criteria were studied: Keneth Jones, described by Stegen et al.(1), World Health Organization(2) and Keith Edwards, described by Crofton et al.(3). Results: Among the criteria studied, those of Keith Edwards showed the best sensitivity and specificity, with 84% and 97%, respectively. WHO criterion showed a 100% specificity but zero sensitivity. Keneth Jones criterion showed 56% sensitivity and a 94% specificity. Conclusion: In poor areas, such as that of this study, where easy-to-handle operational methods are required, the criteria studied proved to be adequate to detect tuberculosis cases in children in contact with tuberculous adults. Those of Keith Edwards were the most useful although some changes are necessary for a better adaptation to Brazil.

Keywords: Tuberculosis, diagnosis. Child. Community health care centers.