To elucidate the role of leptin in acute systemic inflammation, we investigated how its infusion at low, physiologically relevant doses affects the responses to bacterial lipopolysaccharide (LPS) in rats subjected to 24 h of food deprivation. Leptin was infused subcutaneously (0-20 μg/kg/h) or intracerebroventricularly (0-1 μg/kg/h). Using hypothermia and hypotension as biomarkers of systemic inflammation, we identified the phase extending from 90 to 240 min post-LPS as the most susceptible to modulation by leptin. In this phase, leptin suppressed the rise in plasma TNF-α, and accelerated the recoveries from hypothermia and hypotension. Suppression of TNF-α was not accompanied by changes in other cytokines or prostaglandins. Leptin suppressed TNF-α when infused peripherally, but not when infused into the brain. Importantly, the leptin dose that suppressed TNF-α corresponded to the lowest dose that limited food consumption; this dose elevated plasma leptin within the physiological range (to 5.9 ng/ml). We then conducted in-vitro experiments to investigate whether an action of leptin on macrophages could parallel our in-vivo observations. The results revealed that, when sensitized by food deprivation, LPS-stimulated peritoneal macrophages can be inhibited by leptin at concentrations that are lower than those reported to promote cytokine release. It is concluded that physiological levels of leptin do not exert a pro-inflammatory effect, but rather an anti-inflammatory effect involving selective suppression of TNF-α via an action outside the brain. The mechanism of this effect might involve a previously unrecognized, suppressive action of leptin on macrophage subpopulations sensitized by food deprivation, but future studies are warranted.