This study was designed to investigate the effect of tumor necrosis factor-alpha (TNF) on the control of food intake in rat. The specific aims were: a) to evaluate the effects of central (ICV) or peripheral (IP) microdoses of TNF on food intake; b) to show that the TNF-induced anorexia results from a direct action of the mediator on the central nervous system; c) to demonstrate that the anorexic activity of TNF is not due to nonspecific malaise. In the first experiment, ICV administration (0.5-4.0 micrograms) of recombinant-murine tumor necrosis factor-alpha (rmTNF) significantly reduced food intake in a dose-dependent manner. The maximal effect (-66%) was observed 5 h after a 4.0 micrograms dose. In contrast, rm TNF did not affect feeding behavior when injected IP, indicating that the anorexic activity was centrally mediated. The estimated DE50 was 2.0 micrograms/rat by the ICV route. In the second experiment, the place conditioning paradigm was used to show that TNF administered ICV at 2.0 micrograms/rat did not induce aversive or deleterious effects as compared to naltrexone given IP at the equi-anorectic dose 5.0 mg/kg. It was concluded that TNF reduces food intake in rats by a direct action on the central nervous system.