There is significant evidence that the macrophage plays a critical role in the host's defense against neoplasia. Tumor-necrosis factor was recognized by Carswell et al. during a study of the antitumor activity of serum from mice infected with BCG and subsequently injected with endotoxin. The same procedure was applied to rabbits in order to obtain serum containing tumor-necrosis factor (TNF). Sera from these mice and rabbits contained a factor that induced hemorrhagic necrosis of certain mouse sarcomas in vivo and had cytotoxic effects on mouse and human tumor cells in vitro. Sera from mice and rabbits singly treated with BCG or endotoxin did not have these properties. Other agents such as C. parvum, OK-432, lentinan or zymosan, that cause hyperplasia of reticuloendothelial system and increase sensitivity to endotoxin lethality, could substitute for BCG in priming for TNF release. However, the use of P. acnes as a priming agent was the most effective and lipopolysaccharide from gram-negative bacteria appeared to be unique in its ability to elicit TNF release. TNF is a protein with a molecular weight, ranging from 40,000 to 60,000 that has both tumor necrotizing activity in vitro and tumor killing activity in vitro. It is relatively stable to heating at up 70 degrees C. This result indicated that both in vitro and in vitro activities of mouse and rabbit TNF are a property of one and the same molecule. TNF is thought to be produced by macrophage and is distinguished from the other know macrophage products in serum containing TNF. TNF is cytotoxic to several but not all tumor cell lines. Its most interesting feature is that it reportedly dose not affect any non-transformed cell types, implying that it somehow recognizes transformed cells.