Résumé : In this study, the distribution of Puumala hantavirus (PUUV) infection in local bank vole Myodes glareolus populations in an area with low human PUUV infection (nephropathia epidemica [NE]) incidence in northern Belgium was monitored for 2 consecutive years. Bank voles were trapped in preferred habitat and tested for anti-PUUV IgG. Infection data were related to individual bank vole features, population demography, and environmental variables. Rare occurrence of PUUV infection was found and PUUV prevalence was low compared with data from the high NE incidence area in southern Belgium. Small-scale climatic differences seemed to play a role in PUUV occurrence, vegetation index and deciduous forest patch size both influenced PUUV prevalence and number of infected voles in a positive way. The data suggested a density threshold in vole populations below which PUUV infection does not occur. This threshold may vary between years, but the abundance of bank voles does not seem to affect the degree of PUUV seroprevalence further. We found indications for a dilution effect on PUUV prevalence, dependent on the relative proportion of nonhost wood mice Apodemus sylvaticus in a study site. In conclusion, we regard the combination of a dilution effect, a possible threshold density that depends on local conditions, and a higher fragmentation of suitable bank vole habitat in our study area as plausible explanations for the sparse occurrence of PUUV infection and low prevalence detected. Thus, beside human activity patterns, local environmental conditions and rodent community structure are also likely to play a role in determining PUUV infection risk for humans.