BCG revaccination is still used in some tuberculosis endemic countries. Until now, the little evidence available suggested that BCG revaccination confers very limited additional protection, although there was no information on whether protection depends on the setting and age of revaccination, or if protection increases with time since vaccination. Here we report on an extended follow up of the BCG-REVAC trial, a cluster randomised trial conducted in the Brazilian cities Salvador and Manaus including over 200,000 children aged 7–14 years aimed to evaluate the efficacy of BCG revaccination in children who had received neonatal BCG vaccination. With the extended follow-up (9 years) and the additional cases accrued we now have enough power to report vaccine efficacy separately for the two cities (with different distances from Equator and presumably different prevalence of non-tuberculosis mycobacteria), and by age at vaccination and clinical form. The overall vaccine efficacy was 12% (−2 to 24%) as compared to 9% (−16 to 29%) for the 5-year follow up. Vaccine efficacy was higher in Salvador (19%, 3 to 33%) than in Manaus (1%, −27 to 27%) with the highest vaccine efficacy in children from Salvador aged <11 years at revaccination (33%, 3 to 54%). The findings are in line with the hypothesis that BCG vaccination offers higher efficacy in low NTMb prevalence, and show that revaccination with BCG can offer weak protection in selected subgroups.