Parallel monthly surveys of children aged 6-71 months were conducted in the Muheza district of Tanzania. The aim was to compare highland villages, where the mean, annual entomological inoculation rate (EIR) for malaria is 34 and mean annual prevalences of parasitaemia range from 33%-76%, with culturally similar villages of the lowlands, where the mean EIR is 405 and prevalences of parasitaemia range from 80%-84%. The total survey population could be divided into six geographical subgroups, which can be arranged in order of increasing prevalence of parasitaemia. The prevalences of dense parasitaemia, of febrile malaria, and of anaemia all increased in the same order across this series of groups, the trends being statistically significant. The results of previous studies have indicated a paradoxical effect whereby children in regions with a lower exposure to malarial infection suffer, in the long term, a higher incidence of severe attacks of malaria. In the present study there was no sign of any such paradoxical inverse relationship between the level of exposure and the prevalence of malarial illness or anaemia. However, child mortality rates are similar in the highlands and lowlands, as are the median ages of children admitted to hospital. Overall, the present findings indicate that, for the populations studied, an artificial reduction in EIR would be beneficial, even in the long term, with regard to the chronic effects of malaria. This does not necessarily conflict with previous studies reporting opposite conclusions with regard to the incidence of severe,acute effects.