n human beings, infection with the tubercle bacillus usually results in inapparent disease, recognized only by a positive tuberculin reaction. However, about 10% of tuberculin-positive people develop clinically active tuberculosis (TB). The Lurie tubercle count method is probably the most accurate way to measure a vaccine's ability to prevent such clinical disease. Yet, it is rarely used. Briefly, vaccinated and control rabbits are infected by aerosol with a known quantity of virulent human-type tubercle bacilli (strain H37Rv). [Human-type bacilli are not fully virulent for rabbits.] Five weeks later, the rabbits are sacrificed, and counts are made of the number of grossly visible primary tubercles in their lungs. The best vaccines cause the greatest reduction in the number of such visible tubercles. This report describes the method, and the immunologic mechanisms involved. It also suggests how the method can be used to test TB vaccines in both mice and guinea pigs, as well as in rabbits.