There is general agreement that BCG vaccination with a potent strain, when given to previously uninfected subjects, is highly effective in preventing the development of tuberculosis among them. This ‘direct’ effect may be measured in practice in terms of the proportion of cases prevented in the age-groups in which the vaccination has been made. It is also claimed that mass BCG vaccination — especially at school-leaving age — can be expected to yield benefits not only directly, but also indirectly, by breaking the chain of transmission and so preventing the development of tuberculosis in unvaccinated subjects. This ‘indirect’ effect may be measured, by analogy, in terms of the reduction in the numbers of cases in the age groups in which no vaccinations have been performed (the older age groups, and also young children, if BCG vaccination is given, for instance, to school-leavers). The indirect effect will be observed in terms of changes in both smear-positive and smear-negative cases. However, because smear-positive cases are the principal sources of infection, it is more meaningful to measure the indirect effect in terms of the reduction in the numbers of smear-positive (infectious) cases. The present paper falls into three parts: 1. I. A comparison of the trends in the incidence of infectious tuberculosis in Norway, Denmark and The Netherlands. The study confirms that the substantial direct effect of BCG vaccination on the total tuberculosis incidence, which was shown by Bjartveit and Waaler, also applies to the incidence of infectious tuberculosis. 2. II. A comparison of the trends in the incidence of infectious tuberculosis in the three countries and in different age groups over 30, and in the incidence of all forms of tuberculosis in children, in whom BCG vaccination has not been used. These comparisons suggest that the indirect effect of BCG on infectious cases of tuberculosis in persons aged 30 years and over, and on all forms of tuberculosis in children, in whom BCG vaccination has not been used, is not readily detectable, and so may not be large. However, the trends in tuberculosis incidence measure the total effects of all the policies for control used in the three countries under study. The comparisons between the three countries can therefore not isolate the ‘pure’ direct and indirect effects of BCG from the effects of other differences between the control policies. For this reason a theoretical study of the problem has been made. 3. III. A theoretical examination of the maximum likely indirect effect of a mass BCG vaccination policy in diminishing the prevalence of sources of infection in the population. This shows that the effect of BCG in preventing smear-positive cases in developing countries (if a constant risk of infection of about 3 per cent is presumed), is between 0.3 per cent and 2.0 per cent per year. This effect depends mainly upon the efficacy of BCG vaccine, the population covered and the duration of protection from BCG vaccination. Thus it seems that BCG vaccination, even if used in a mass-campaign at ages 15–30 years, will not substantially influence the chain of transmission, especially if the risk of tuberculosis infection is high and has not been decreasing. However, it has to be stressed that mass BCG vaccination was introduced for its direct effect and it should be applied whenever it is justified.