To study the ecologic correlates of hantavirus in deer mice (Peromyscus maniculatus), we sampled 114 sites in the Walker River Basin of Nevada and California in 1995-1996. Blood samples were tested for antibody to hantavirus, and a subset of samples was also tested for virus RNA by reverse transcription-polymerase chain reaction. Average prevalence of antibody-positive mice was 17%, with heavier males the most likely to be infected. Antibody prevalence varied within repeatedly sampled sites from 0% to 50% over the course of several months, suggesting possible infection cycles. Although there was no linear correlation between deer mouse density and antibody prevalence on sample sites, more complex relationships between density and prevalence appeared likely. Specifically, infections were less likely where rodent densities were lower than a critical threshold value. However, above this value, density had no effect on prevalence.