This study investigated the epidemiology of uncomplicated falciparum malaria in an area of unstable and seasonal transmission in eastern Sudan. About 90% of malaria morbidity in this region occurs in the months of September to November, and very few malaria cases occur during the intensely arid Sudanese dry season and during years of drought. The malaria situation in the study site, the village of Daraweesh, was analysed during 3 consecutive malaria seasons in 1993-95 during which the 457 inhabitants suffered at total of 436 episodes of falciparum malaria. Using an Andersen-Gill proportional hazard model for recurrent events stratified by family, we have calculated the relative hazard for clinical malaria episodes by age, sex, haemoglobin genotype, blood type and infection in the previous season. The malaria risk was significantly lower in individuals aged 20-88 years than in the 5-19 years age-group. The relative protection due to adulthood varied between seasons (relative risk, RR, 0x34 to 0x67). Serological data were not consistent with the hypothesis that the age difference in incidence was due to differences in exposure. During the 1993 season the malaria incidence in males was lower than in females (RR = 0x75), during the 1994 season the incidences were comparable, whereas males had an increased risk of malaria in 1995 (RR = 1x87). The relative risk in individuals carrying the haemoglobin AS genotype compared to homozygous AA individuals was 0x57.