Human plague risks (Yersinia pestis infection) are greatest when epizootics cause high mortality among this bacterium's natural rodent hosts. Therefore, health departments in plague-endemic areas commonly establish animal-based surveillance programs to monitor Y. pestis infection among plague hosts and vectors. The primary objectives of our study were to determine whether passive animal-based plague surveillance samples collected in Colorado from 1991 to 2005 were sampled from high human plague risk areas and whether these samples provided information useful for predicting human plague case locations. By comparing locations of plague-positive animal samples with a previously constructed GIS-based plague risk model, we determined that the majority of plague-positive Gunnison's prairie dogs (100%) and non-prairie dog sciurids (85.82%), and moderately high percentages of sigmodontine rodents (71.4%), domestic cats (69.3%), coyotes (62.9%), and domestic dogs (62.5%) were recovered within 1 km of the nearest area posing high peridomestic risk to humans. In contrast, the majority of white-tailed prairie dog (66.7%), leporid (cottontailed and jack rabbits) (71.4%), and black-tailed prairie dog (93.0%) samples originated more than 1 km from the nearest human risk habitat. Plague-positive animals or their fleas were rarely (one of 19 cases) collected within 2 km of a case exposure site during the 24 months preceding the dates of illness onset for these cases. Low spatial accuracy for identifying epizootic activity prior to human plague cases suggested that other mammalian species or their fleas are likely more important sources of human infection in high plague risk areas. To address this issue, epidemiological observations and multi-locus variable number tandem repeat analyses (MLVA) were used to preliminarily identify chipmunks as an under-sampled, but potentially important, species for human plague risk in Colorado.