A total of 736 outpatients diagnosed as having malaria using clinical criteria at a health centre in a highly endemic area of Papua New Guinea were investigated parasitologically. Plasmodium falciparum-attributable fractions were determined using a logistic regression model to compare parasite densities in cases with those of healthy individuals in community surveys. Thirty-seven percent of presumptive cases were found to have raised P. falciparum parasitaemia. This corresponds to an average reporting rate for the population of 0.53 attributable episodes per annum. Whilst the maximum prevalence of parasitaemia in the community was in children aged 5-9 years, the maximum age-specific incidence of attributable cases at the outpatient clinic was 2 cases per annum in the 2- to 4-year-old age group. The procedure for estimating attributable fractions makes it possible to compare morbidity rates between age groups, and to examine how the relationship between morbidity risk and parasite density changes with age, without diagnosing individual episodes. The average tolerance of parasites in an age group was measured by considering the level of parasitaemia associated with a given risk of malaria-attributable morbidity. In contrast to anti-parasite immunity, tolerance of parasites declines with age since at parasite isodensity the probability of being symptomatic increases with age.