In areas where the agent of Lyme disease is intensely enzootic, the mouse reservoirs may be universally infected. Because a large proportion of the vector tick population appears to feed upon these hosts, the prevalence of infection in the vectors should approach 100%. However, infection in host-seeking nymphal ticks in nature rarely exceeds 40%. To help reconcile this apparent paradox, we examined whether estimates of prevalence might differ if we did not assume that infected ticks are randomly or uniformly distributed within a site. Nymphal Ixodes dammini were collected by dragging a series of 10-meter replicates within an intensely enzootic site. Estimates of the prevalence of spirochetal infection, based upon the monthly means of individual 10-meter collections, were then compared with estimates derived by pooling all samples. Host-seeking ticks tended to cluster. The Lyme disease spirochete was present in 15.6% of 469 pooled ticks. When the prevalence estimate was based solely on ticks in clusters that contained one or more infected ticks, however, at least 50% of the ticks were infected. We conclude that nymphal deer ticks infected by Lyme disease spirochetes tend to aggregate spatially in nature, and that prevalence estimates based upon a mean value for pools may be misleading.