This paper describes the design and baseline results of a large and simple randomized controlled trial of the protection against tuberculosis of a dose of Bacillus Calmette Guerin (BCG) vaccination given to school children in a population with a high coverage of neonatal BCG (The Brazilian BCG-REVAC trial). The study started in 1996 and is a pair-matched and stratified-cluster randomized controlled trial with no placebo. The study population consists of children aged 7-14 years enrolled in 763 state schools from the cities of Salvador and Manaus, Brazil. Schools were the unit of randomization. Identifying information was collected for 354,708 school children. The final study population, after exclusions on the basis of age, BCG scar readings and absence from school on the day of the study visit, consists of 242,401 children, of whom 125,403 are in intervention schools. Follow-up relies on ascertainment of cases diagnosed at the health services and notified to the tuberculosis control program surveillance system. Blindness is guaranteed during linkage and validation of cases. Analysis is planned for the next 12 months, where efficacy will be estimated by calculating incidence of tuberculosis in the vaccine and control groups, taking into consideration the cluster design. The intervention studied, a second BCG vaccination, is widely used, although the World Health Organization does not recommend it on the basis of absence of evidence of protection or lack of protection. The results of the trial will make it possible for BCG revaccination practice to be informed by evidence. This is an example of a large simple and relatively inexpensive effectiveness trial, resulting from good collaboration between academia and health and education services enabling developing countries to define policies that are relevant for their reality.