Cerebral malaria in man and in mice is the consequence of a cascade of events, involving the production of toxins by the parasite and cytokines by the host, and eventually leading to the amplification of the expression of the receptors for cytoadherence on brain capillary endothelial cells. Variations in the intrinsic characteristics of parasite isolates or the genetic make-up of the host and the degree of antimalarial immunity can modulate this sequence of events. A working hypothesis is proposed in which two features of the parasite, the ability to cytoadhere and to produce toxins, are clearly dissociated and where the amplification of cytoadherence receptors is considered crucial. This hypothesis, illustrated by new data from human malaria and rodent models, suggests that cerebral malaria may occur when these features occur together during an infection, while not necessarily within the same parasite clone.