Suter, Ullman, and Hoffman (1) were the first to report that mice infected with an attenuated strain of Mycobacterium tuberculosis (BCG) are more susceptible than control animals to the lethal effects of bacterial endotoxins derived from Gram-negative organisms. This was confirmed by Howard et al. (2) who also showed that increased activity of the reticuloendothelial system (RES) exists at a time when the animals are most sensitive to endotoxin. This finding had not been anticipated since the RES is believed from other experiments to play an important role in an animal's response to endotoxin. "Blockade" of the RES by injections of colloidal agents sensitizes animals to endotoxin (3, 4) while induced tolerance to endotoxin is lost following similar treatment (5, 6). More recently, Suter and Kirsanow (7) elaborated the conditions necessary for the development of hyperreactivity toward endotoxin in BCG-infected animals. Infection with living mycobacteria is more effective than injection of an equivalent number of dead organisms. Maximum sensitivity develops after 7 to 9 days and persists for some time thereafter. The reaction is non-specific; i.e., hyperreactivity occurs with endotoxins derived from different species of Gram-negative bacteria.Our interest in this effect of BCG was prompted by the desire to increase the sensitivity of the urinary nitrogen excretion assay for endotoxin (8,9). It soon became apparent, however, that BCG-infected mice respond to endotoxin in a manner quite distinct from those previously studied in our laboratory: This report summarizes the results on which this conclusion is based.