The present study tested the hypothesis that the cytokine tumor necrosis factor-alpha (TNF-alpha) is an important central nervous system mediator of the rat hypothalamo-pituitary-adrenal (HPA) axis response to the i.v. administration of lipopolysaccharide (LPS; 5 microg/kg). LPS produced a rapid (within 30 min) rise in plasma TNF-alpha levels, which preceded elevations in plasma ACTH (commencing at 45 min). Despite a lack of detectable TNF-alpha biological activity in the brain 30 min to 2 h after LPS administration, intracerebroventricular (i.c.v.) pretreatment (-20 h) with 5 microl anti-TNF-alpha antiserum significantly delayed the onset of the plasma ACTH response to LPS, suggesting that TNF-alpha acts within the brain. However, we also noted that the i.c.v. infusion of anti-TNF-alpha 20 h earlier produced experimentally significant concentrations of the same anti-TNF-alpha antibodies in systemic blood. This suggested the possibility that the effect of this antiserum was due to its leakage to the periphery. Indeed, 5 microl anti-TNF-alpha administered i.v. at -20 h produced an inhibition of the ACTH response to LPS that was temporally and quantitatively similar to that produced by i.c.v. anti-TNF-alpha. Intracerebroventricular administration of anti-TNF-alpha immediately before LPS produced only low systemic blood levels of corresponding anti-TNF-alpha antibodies and did not significantly alter the plasma ACTH response, whereas i.v. administration of anti-TNF-alpha immediately before LPS was clearly effective. Collectively, these results show that 1) biologically active levels of TNF-alpha in systemic plasma and the ensuing ACTH responses to LPS were always temporally and qualitatively related; and 2) even though i.c.v. administration of anti-TNF-alpha could inhibit the HPA axis response to LPS, this was apparent only when substantial amounts of anti-TNF-alpha antibodies had reached systemic blood. We, therefore, conclude that at the dose of LPS used in this study (5 microg/kg), TNF-alpha is an important mediator of the HPA axis response to LPS by an action within the periphery, but probably not within the brain.