Evidence is accumulating that the illness and pathology observed in malaria are not caused directly by parasite products, but by normal components of the immune response, mainly monokines such as tumor necrosis factor (TNF), produced in excess. These mediators are released from the host's monocytes and macrophages, apparently in response to stimulation by parasite products. Recombinant TNF, if injected into a range of animal species or into tumour patients, is demonstrably toxic, giving rise to changes typical of acute malaria, and several groups have detected circulating TNF in serum from patients acutely ill with malaria. The short serum clearance time of TNF and TNF tolerance have to be considered when interpreting such data. Current studies indicate that some malarial antigens, in the absence of lipopolysaccharide, can trigger release of TNF. This and other monokines could contribute to cerebral malaria in at least 2 ways: by increasing thrombospondin secretion, and hence favouring local sequestration of knob-bearing parasitized red cells, and, as has been demonstrated in clinical trials in tumour patients, by causing neurological symptoms directly. In addition, it seems that TNF does not act alone, but as part of an interdependent synergizing network of polypeptide mediators. These evidently act together to induce secretion of other cell products, such as platelet-activating factor, prostaglandins, reactive oxygen species and procoagulant activity, that actually cause illness, biochemical change and tissue damage. Understanding these processes should lead to a range of new therapeutic interventions.