This article reports the results of a national tuberculin skin test survey of childhood age group. The survey period was from december 1991 to june 1994. For the calculation of annual risk of tuberculous infection, 1544 schoolchildren, aged 6 to 10 years old, without scare related to BCG, were included. The antigen used was tuberculin PPD Mérieux; in order to define a mode of positivity, this tuberculin was first tested with 250 confirmed pulmonary tuberculous patients: the mode was 16mm. With the hypothesis of a cut-off point of positivity at 14 mm, the prevalence of tuberculous infection was found at 9.6% (with a confidence interval of 1.6%); then, the annual risk of tuberculous infection was calculated at 1.21% (from 1% to 1.42%). With the hypothesis of a mode at 16 mm, the prevalence of tuberculous infection was found at 10.2% (with a confidence interval of 2.3%); then, the annual risk of tuberculous infection was calculated at 1.29% (from 0.97% to 1.59%). Considering separately two age groups, the annual risk was 1.25% for 6-8 years old children and 1.32% for 9-10 years old children. In conclusion, the authors stress the important constraints in the achievement of such a survey in developing countries. Though the difficulties that arised, the standardised methodology used in this survey gives reliable information. These results could be compared with those of future surveys using the same methodological approach.