Data on prairie dog densities, flea abundance on prairie dogs, and plague epizootics in Montana and Utah, USA, 2003-2005. Prairie dog species (PDspecies in the data file) included black-tailed prairie dogs (PDs) (BTPD, Cynomys ludovicianus) in north-central Montana, white-tailed PDs (WTPD, Cynomys leucurus) in eastern Utah, and Utah PDs (UPD, Cynomys parvidens) in southwestern Utah. Field research was completed by the U.S. Geological Survey, Fort Collins Science Center, and colleagues. We used summertime visual counts as an index to PD densities (Pddensity in the data file). For each plot, we counted PDs using binoculars and/or spotting scopes from a single location outside the plot that gave the best view of the entire plot and repeated these counts on three (usually consecutive) days. We began counts just after sunrise and continued to conduct repeated systematic scans of the plot until the counts declined to about half the peak number (usually by late morning as PDs went below ground for their typical mid-day break). We converted the counts to density estimates (counts per hectare [ha]).The estimate we used to calculate density was the highest count obtained from a plot for the 3 days within a given year. We analyzed data from colonies experiencing a plague epizootic during this particular study (with an epizootic defined as greater than or equal to 90% decline in PD density). We indexed annual population change (PDpopchgProportion in the data file) by subtracting the count density estimate of the year before a plague epizootic (t1) from the density estimate during an epizootic (t2) for each plot, and dividing that by the density estimate from t1 to summarize population change as a proportionate change. We evaluated the correlation between PD population change and PD density in year t1, because negative plague-effects and the intensity of population decline may be greatest when PD densities are high in year t1 (a potential "density dependent" phenomenon discussed in a wide range of literature on disease ecology). We also evaluated the correlation between PD population change and flea abundance in year t1, because rates of plague transmission and, therefore, PD mortality are expected to increase with increasing flea densities. To assess flea abundance (PDfleas in the data file), we combed live-trapped PDs and counted the number of fleas on each PD. The PDs were live-trapped, individually marked with ear tags, and combed as thoroughly as possible for 30 seconds (s) to collect fleas. Prairie dogs were allowed to recover from anesthesia and released at their trapping locations. For each plot and year, we used the average value of flea counts (defined as flea abundance).