For the diagnosis of rickettsial diseases, though there are agglutination and complement fixation tests with the use of pure rickettsial suspension, the Weil-Felix test is still of high value for its technical simplicity and specificity. Even in the case of Tsutsugamushi disease, Fletcher, Wolf, Felix and Kawamura applied the Weil-Felix reaction with the use of Proteus OXK bacilli, and since then this reaction has been discussed in various ways, but the true antibody titer seems not to be obtained because of the insufficient consideration of the antigen concentration. Especially when the variation in the production of antibody is discussed on the basis of antibody titer, there apt to be misinterpretations. By the Weil-Felix reaction with the use of Proteus OXK dacilli suspension of 50% agglutination titer as the antigen, the author examined the variation of the antibody production in the sera of the test animals inoculated with rickettsiae, in parallel with the case in which the antigen was selected at random in the conventional way. As the result, it was found that when the antigen of 50% agglutination titer was used, no variation was shown in the antibody titer, irrespective of the sites of inoculation, the number of times of inoculation, life or death of rickettsiae, while, in the case of antigen selected at random, variable titers were yielded. From these data, the author believes that the antigen concentration suitable for she Weil-Felix reaction has been found.