Background: There is a low incidence of malaria in Iquitos, Peru, suburbs detected by passive case-detection. This low incidence might be attributable to infections clustered in some households/regions and/or undetected asymptomatic infections. Methods: Passive case-detection (PCD) during the malaria season (February-July) and an active case-detection (ACD) community-wide survey (March) surveyed 1,907 persons. Each month, April-July, 100-metre at-risk zones were defined by location of Plasmodium falciparum infections in the previous month. Longitudinal ACD and PCD (ACP+PCD) occurred within at-risk zones, where 137 houses (573 persons) were randomly selected as sentinels, each with one month of weekly active sampling. Entomological captures were conducted in the sentinel houses. Results: The PCD incidence was 0.03 P. falciparum and 0.22 Plasmodium vivax infections/person/ malaria-season. However, the ACD+PCD prevalence was 0.13 and 0.39, respectively. One explanation for this 4.33 and 1.77-fold increase, respectively, was infection clustering within at-risk zones and contiguous households. Clustering makes PCD, generalized to the entire population, artificially low. Another attributable-factor was that only 41% and 24% of the P. falciparum and P. vivax infections were associated with fever and 80% of the asymptomatic infections had low-density or absent parasitaemias the following week. After accounting for asymptomatic infections, a 2.6-fold increase in ACD+PCD versus PCD was attributable to clustered transmission in at-risk zones. Conclusion: Even in low transmission, there are frequent highly-clustered asymptomatic infections, making PCD an inadequate measure of incidence. These findings support a strategy of concentrating ACD and insecticide campaigns in houses adjacent to houses were malaria was detected one month prior. © 2005 Branch et al; licensee BioMed Central Ltd.