It is proposed that case-control studies might be used to assess the effect on the incidence of tuberculosis of the mass BCG vaccination campaigns that were conducted in many countries in Africa and Asia from the 1950s onwards. The cost of such studies would be relatively small and they could be completed within a year or two. The results would assist countries in planning their future strategies for tuberculosis control and information on the variation in the protective effect of BCG in a number of different areas in less developed countries may give clues to the reasons for such variation. It is suggested that case-control studies should be designed to include about 200 cases and 3 or 4 times as many controls. The cases should be selected for age such that at the time of the mass vaccination campaign they will have been in an age group in which 20% or less were tuberculin positive. The same approach may be used to assess the effectiveness of giving BCG at birth, provided that this policy has been adopted for some years, coverage has been good (about 50% to 80%) and there are reasonable facilities for diagnosing tuberculosis in children.