SmmaryAdministration of either tumour necrosis factor alpha or 16,16-dimethylprostaglandin E2 (PGE2) to female NMRI mice caused a decrease in body weight accompanied by a reduction in both food and water intake and a decrease in carcass water content. A single injection of TNF-2 caused an enhanced production of PGE2 by spleen cells from treated animals, that was significant within I h of treatment, and persisted until at least 6h. These results suggest that the anorectic effect of TNF-z may be mediated by a prostaglandin intermediate. Indomethacin (lOmgkg-) administered 2h before TNF-a (7.5 x 107 U kg-') caused a significant reduction in the extent of weight loss and inhibited PgE2 production. Administration of indomethacin 0.5-1.5 h before the TNF-z had no significant effect on loss of body weight, but still inhibited PgE2 production. Also PgE2 production was still enhanced in response to TNF-2 administered chronically, despite the inability of prolonged TNF-z administration to produce continued loss of body weight. These results suggest that prostaglandins are not involved in the anorectic effect of TNF-z.The catabolic states associated with infection or endotoxemia have been attributed to the production by phagocytic cells of soluble proteins such as interleukin 1 and cachectin (Rouzer & Cerami, 1980;Moldawer et al., 1987a;Cerami et al., 1985). A high degree of homology has been shown to exist between the N-terminal sequence of mouse cachectin and the N-terminal sequence for human tumour necrosis factor-alpha (TNF-x) and the catabolic states have been extended to include also cancer-associated cachexia (Beutler & Cerami, 1986). Severe weight loss and increased mortality have been observed in mice bearing transgenic tumours that persistently secrete human cachectin (Oliff et al., 1987) and chronic administration of sublethal doses of TNF-x to rats caused anorexia, weight loss, depletion of body lipid and protein, a reduction of red blood cell mass, leukocytosis and tissue inflammation (Tracey et al., 1988). However, in a number of studies (Stovroff et al., 1988;) administration of TNF-x caused a loss in body weight accompanied by a drop in food and water intake, which was only apparent over the first 24h, after which animals became resistant to subsequent dosing. In addition TNF-x has not been detectable in the serum of patients with clinical cancer cachexia (Socher et al., 1988) and in clinical tnrals of recombinant human TNF-x there was no clinical evidence of accelerated cachexia, although anorexia was present during administration (Sherman et al., 1988). Also the effects on host metabolism produced by TNF-:x appear to differ from that produced by a cachexiainducing tumour .The toxic and metabolic effects of TNF-:z can be blocked by a single injection of the cyclo-oxygenase inhibitors indomethacin or ibuprofen before the TNF-x treatment (Kettlehut et al., 1987). This suggests that some of the effects of TNF-x may be mediated through a prostaglandin intermediate in analogy with septic shock where l...