In mice, tumor necrosis factor (TNF) displays a selective species specificity. In contrast to murine TNF (mTNF), human TNF (hTNF) only induces lethality at extremely high doses of about 500 μg/mouse, whereas it still has a powerful antitumor activity in combination with interferon‐γ. The observation that hTNF does not interact with the p75 mTNF receptor seemed to provide a plausible explanation for these species‐specific biological effects. Experiments in TNF receptor knockout mice and tests with hTNF muteins in baboons did not, however, support this hypothesis. We here show that an mTNF mutein selective for the p55 mTNF receptor induces lethality in a manner comparable to wild‐type mTNF, and conclude that other differences between hTNF and mTNF must account for the reduced lethality of hTNF. Pharmacokinetics showed that hTNF is cleared much faster than mTNF or the mTNF mutein used. In contrast to the hardly lethal effect(s) of a bolus administration of hTNF, fractionated repetitive administration of the same total hTNF dose induced lethality. This suggests that prolonged exposure rather than peak levels determine the lethal effects of hTNF in mice. Experiments with receptor and ligand knockouts demonstrated that the difference in pharmacokinetics is independent of an interaction with (soluble) TNF receptor, TNF‐induced effects or induction of endogenous TNF. These results show that manipulation of the clearance rate of TNF may broaden the therapeutic range of systemic treatments with TNF.