There are several new types of vaccines that have become potential candidates for the prevention of tuberculosis. While these have been tested in the conventional animal models, less emphasis has been placed as yet on the development of safety testing procedures. This is particularly important if we are ever going to use such vaccines in a therapeutic or post-exposure mode, in which the risk of severe “Koch-like” reactions may be high. A further element that badly needs more research is the state of immunity already preexisting in the infected host and how vaccine induced immunity may interact with this.