During a four-year study with direct latex agglutination tuberculosis, many properties of this test were clarified. A serologically active tuberculoprotein antigen was prepared that is stable in the lyophilized state. In order to eliminate spontaneous agglutination, which results in false positive readings, the test was set up in two series of tubes with non-sensitized and sensitized latex. The test was applied to serum specimens from 462 patients. Of 197 patients with active tuberculosis, agglutination occurred with the antigen in 96.9 per cent. After deducting the significant spontaneous agglutinations that occurred in 6.6 per cent, a final result of 90.3 per cent was observed. Among 84 cases of inactive tuberculosis, antibodies were demonstrated in 38.5 per cent, and in 23.7 per cent of 72 patients with cutaneous reactions to tuberculin and no demonstrable disease. Among 33 patients with no reaction to tuberculin and no manifest pulmonary disease, one serum specimen gave an agglutination with a titer of 8. However, of 109 hospital and clinic patients with different diagnostic conditions, nonspecific agglutination occurred in 12 per cent. Repeated tests from patients with stable conditions did not reveal variations in the agglutination titer. In progressive tuberculous disease, the agglutination titer usually increased, and gradually decreased with improvement of the disease. When inactive status was reached, the agglutination titer returned to normal in approximately one third of the patients. The test contributes additional data of aid in differentiation of active and inactive tuberculosis.