In guinea pigs, dose-dependent febrile responses were induced by injection of a high (100 g/kg) or a low (10 g/kg) dose of bacterial lipopolysaccharide (LPS) into artificial subcutaneously implanted Teflon chambers. Both LPS doses further induced a pronounced formation of prostaglandin E 2 (PGE2) at the site of localized subcutaneous inflammation. Administration of diclofenac, a nonselective cyclooxygenase (COX) inhibitor, at different doses (5, 50, 500, or 5,000 g/kg) attenuated or abrogated LPS-induced fever and inhibited LPS-induced local PGE2 formation (5 or 500 g/kg diclofenac). Even the lowest dose of diclofenac (5 g/kg) attenuated fever in response to 10 g/kg LPS, but only when administered directly into the subcutaneous chamber, and not into the site contralateral to the chamber. This observation indicated that a localized formation of PGE 2 at the site of inflammation mediated a portion of the febrile response, which was induced by injection of 10 g/kg LPS into the subcutaneous chamber. Further support for this hypothesis derived from the observation that we failed to detect elevated amounts of COX-2 mRNA in the brain of guinea pigs injected subcutaneously with 10 g/kg LPS, whereas subcutaneous injections of 100 g/kg LPS, as well as systemic injections of LPS (intra-arterial or intraperitoneal routes), readily caused expression of the COX-2 gene in the guinea pig brain, as demonstrated by in situ hybridization. Therefore, fever in response to subcutaneous injection of 10 g/kg LPS may, in part, have been evoked by a neural, rather than a humoral, pathway from the local site of inflammation to the brain. lipopolysaccharide; febrile response; prostaglandin E 2; cyclooxygenase-2; immune system-to-brain communication MOST EXPERIMENTAL STUDIES employing bacterial lipopolysaccharide (LPS) to induce fever use a systemic route of administration, such as intraperitoneal, intravenous, or intra-arterial injections of the pyrogen. These treatments result in a generalized systemic inflammatory response, which is accompanied by increased measurable amounts of proinflammatory cytokines and LPS in the circulation (10,14,17,28,32). To study a cytokine release within a local area of inflammation, experimental models have been introduced in which LPS is administered locally, rather than systemically, into a subcutaneous air pouch in rats (8,22,23) or a subcutaneously implanted artificial Teflon chamber in guinea pigs (29,30,34). LPS that is injected into the air pouch (8) or the subcutaneous chamber (29) is not measurable in the systemic circulation 1 h after its injection. This observation does, however, not exclude the possibility that a small fraction of the injected LPS might have leaked from the air pouch or the chamber immediately after the injection procedure. Among the pyrogenic cytokines (TNF, IL-1, or IL-6) that are produced within the site of localized subcutaneous inflammation, IL-6 is the only putative endogenous pyrogen that enters the circulation in significant quantities. It has therefore been proposed that...