1. An analytical study has been made of the mechanism of natural bactericidal action by the serum of various animals (ox, sheep, horse, rabbit, guinea-pig, rat, man) towards certain organisms (B. typhosus, B. dysenteriae Shiga, B. proteus, V. cholerae) exhibiting the maximum reactivity to this effect.2. Serum-complement has no bactericidal action per se, and an antibodylike agent invariably acts as an intermediary agent, “sensitising” the particular organism to the action of the complement and capable of being “absorbed” by it from serum at 0° C.3. This sensitising agent is stable at 55° C. but labile at 60°–65° C. In this respect it resembles natural haemolysins and agglutinins, but contrasts with the more stable immune antibodies and the more labile natural complement-fixing antibodies (for bacterial antigens). It is resident mainly in the carbonic-acid-insoluble fraction of the serum. It is present in the serum of young animals before certain other natural antibodies have developed.4. Absorption tests demonstrate the high degree of specificity of these natural bactericidal antibodies for particular bacteria.5. A non-specific extracellular substance occurs in bacterial cultures which may neutralise or inhibit these antibodies, and interfere with their sensitising action even at 0° C.6. This substance is liberated in large amount in cultures heated at high temperatures (120° C). It can be removed by repeated washing of growths in saline solution. It may inactivate a bactericidal antibody in heated serum, though not in fresh unheated serum, and may inactivate a particular antibody in the serum of one animal species but not in another. Strains of bacteria vary in their production of this substance.7. The observations submitted in this paper, correlated with previous studies of natural antibodies by the authors and others, indicate that immune antibodies have their precursors specifically differentiated in the serum of normal animals and that, in general, immune antibodies are not substances formed de novo.