There is conflicting evidence as to whether Bacillus Calmette-Guérin (BCG) vaccination offers protection against Mycobacterium tuberculosis infection ascertained by a positive tuberculin skin test (TST). We investigated the association between BCG vaccination status and TST results in a set of surveys at increasing TST cut-off values to take cross-reactions to BCG vaccination into account. Secondary analysis of data from three consecutive tuberculin surveys done among schoolchildren in Tanzania between 1990 and 2002. BCG vaccination status was ascertained by the presence of a typical scar. We analyzed data of 277,588 children of whom 77.7% were BCG vaccinated and 8.5% had TST indurations ≥ 15 mm, 5.1% ≥ 17 mm and 2.8% ≥ 19 mm. In the combined analysis, odds ratios for a positive TST were > 1 for children with BCG up to TST cut-off values of 16 mm. For cut-off values > 17 mm crude and adjusted odds ratios were significantly < 1, and decreased with further increasing cut-off values. Using a methodology that makes use of the differences in TST reaction sizes between specific and non-specific responses, we showed that BCG vaccination was associated with reduced prevalence of M. tuberculosis infection as measured by the tuberculin skin test, suggesting a protective effect.