Malaria mortality and morbidity were studied in a rural population of 4000 in the Wosera area, East Sepik Province, Papua New Guinea. Malaria accounted for 4.9% of the 162 deaths investigated by verbal autopsy and for 12.2% of the 49 deaths assessed through medical records. Malaria was the first cause of death in children aged 0.5-4 years. Of the 7795 subjects interviewed and bled during six cross-sectional community-based surveys, children of 1-4 years had the highest malaria-related morbidity. In this age group, point prevalences of fever, fever associated with parasitaemia, and fever plus Plasmodium falciparum (Pf) parasitaemia > or = 10,000 parasites/microliters blood were 5%, 4.1% and 1.5%, respectively. The corresponding figures for adults were 2%, 0.9% and 0.1%, respectively. The calculation of attributable fraction (AF) using a multiple logistic regression model showed that malaria accounted for 0.44 of all fevers in children of 1-4 years and 0.08 of the fevers in adults. Prevalence data derived from the AF estimate were compared with those calculated using different accepted density thresholds. The prevalences which best approximated the results from the logistic regression model were obtained using parasitaemia cut-offs of > or = 1000 Pf parasites/microliter in children aged 1-4 years and adults older than 19 years and of > or = 10,000 parasites/microliter in those aged 5-19 years. Prevalence of fever associated with parasitaemia was highly seasonal, with a peak at the beginning of the wet season. The geographical distribution of malaria morbidity was not uniform. The measurement of malaria-related morbidity, the identification of significant seasonal and local variation as well as the assessment of different methods of defining a clinical episode of Pf malaria are crucial for the design and evaluation of intervention studies, including field trials of antimalarial vaccines.