As indicated by its name, tumor necrosis factor (TNF), cloned in 1985, was originally described as a macrophage-derived endogenous mediator that can induce hemorrhagic necrosis of solid tumors and kill some tumor cell lines in vitro. Unfortunately, its promising use as an anticancer agent was biased by its toxicity, which was clear soon from the first clinical trials with TNF in cancer. Almost at the same time TNF was being developed as an anticancer drug, it became clear that TNF was identical to a mediator responsible for cachexia associated with sepsis, which was termed cachectin. This research led to the finding that TNF is, in fact, the main lethal mediator of sepsis and to the publication of a huge number of articles showing that TNF inhibits the toxic effects of bacterial endotoxins, which are now described as systemic inflammatory response. Although the clinical trials with anti-TNF in sepsis have not been successful thus far, undoubtedly as a result of the complexity of this clinical setting, these studies ultimately led to the identification of TNF as a key inflammatory mediator and to the development of anti-TNF molecules (soluble receptors and antibodies) for important diseases including rheumatoid arthritis and Crohn's disease. On the other side, the mechanisms by which TNF and related molecules induce cell death have been studied in depth, and their knowledge might, in the future, suggest means of improve the therapeutic index of TNF in cancer.