A contact study was undertaken in Seoul to determine the protective effect of the BCG programme in children up to 5 years of age. There were 1993 contact children to 4484 smear-positive patients with pulmonary tuberculosis. 1223 completed the examination, 806 had evidence of BCG vaccination, 417 had not. In total, 129 or 126 children were considered cases of tuberculosis according to radiological/clinical classification or scoring system, respectively. For the unvaccinated the respective numbers of cases were 84 and 80 and for the vaccinated 45 and 46. The data were stratified for factors that could have distorted comparability: age and sex, relationship of index case, feeding habits, room occupancy, treatment history of index case and health centre that diagnosed the index case. Only age was found to have a small effect. Correcting for this, the observed level of protection was 74% with 95% confidence limits of 62% and 82%. It appeared to be the same for all types of disease observed.