Field studies of malaria in endemic areas frequently use the presence or levels of parasitaemia, together with the measurement of fever, as the primary criteria with which to identify cases. However, since malaria cases do not always present with measurable fever, and since asymptomatic parasitaemia occurs, additional episode markers might be useful epidemiological tools. We have measured the C-reactive protein and haptoglobin levels in paediatric patients presenting to a village health post in the Kilombero District in Tanzania and in convalescent sera from the same patients, in order to evaluate these acute-phase reactants as alternative markers of Plasmodium falciparum episodes. Among afebrile patients, C-reactive protein levels were highly correlated with parasite density. High C-reactive protein levels are therefore probably indicative of recent clinical malaria episodes in currently afebrile individuals with high parasite densities. An appropriate case definition for malaria in epidemiological studies in endemic areas might therefore be hyperparasitaemia accompanied by either, or both, measurable fever and raised C-reactive protein levels. This would give less biased estimates of the overall burden of malaria morbidity than does a definition which requires measurable fever. Levels of haptoglobin were highly negatively correlated with parasitaemia, but did not appear to be useful episode markers because this correlation was probably not related to acute morbidity. However, haptoglobin can be useful to assess at community level the impact of interventions on parasitaemia.