The immunity induced in mice by vaccination with living attenuated cultures of tubercle bacilli was measured by two criteria. (a) Increase in survival time of the vaccinated animals after infection with a dose of virulent bacilli sufficient to kill all the unvaccinated controls within 10 to 20 days. (b) Difference in the number of living bacilli recovered from the spleen and lungs of vaccinated and normal animals infected with a small dose of virulent bacilli. The level of immunity induced was found to depend upon the extent of multiplication in vivo of the bacilli used for vaccination. This in turn was conditioned by the degree of attenuation characteristic of the bacterial strain used in the preparation of the vaccine, the amount of vaccine injected, the route of vaccination, and the time interval between vaccination and challenge infection. It was possible to prevent or retard the development of immunity by treating the mice in course of immunization with a drug, isoniazid, capable of interrupting the multiplication in vivo of the bacilli used as vaccine. Although immunity regularly developed and lasted for many weeks when the proper conditions of vaccination were used, the immune response was never sufficient to protect the animals against ultimate death from infection with virulent tubercle bacilli. The prolongation of life in the vaccinated mice was not consequent on a direct bactericidal effect but rather on a retarded or interrupted multiplication of the virulent bacilli in vivo. The quantitative bacteriological techniques used in the present study would appear to be of value for the analysis of certain problems of immunity, and for the appraisal of vaccines and techniques of vaccination.