We collected fleas from black-tailed prairie dogs (Cynomys ludovicianus) on Buffalo Gap National Grassland and Badlands National Park, South Dakota, 2020-2022. We anesthetized trapped prairie dogs (and their fleas) with isoflurane in induction chambers for processing. We combed each individual with a fine-tooth comb for 30 seconds to remove and count fleas. We released all animals, after they were recovered from anesthesia, at the location where they were captured. Total flea counts from individual prairie dogs were considered an index of flea abundance. During 2018-2020, we provided prairie dogs on 21 sites with grain bait, laced with fipronil (0.005% by weight) by placing 1/2 cup or 1/4 cup of grain at each prairie dog burrow opening on the treated sites; 18 non-treated sites functioned as experimental baselines. In 2020–2022, we live-trapped, anesthetized, and combed prairie dogs for fleas. The data release contains two spreadsheets. The first spreadsheet (FipronilNoneComparison.csv) includes data from captures and combings of individual prairie dogs on treated and non-treated sites and lists the following information for each prairie dog: DATE (date of sampling), TREATMENT (fipronil grain or none), and FLEAS (the number of fleas combed from the prairie dog). These data were used to evaluate the effect of fipronil grain on prairie dog fleas. The second spreadsheet (DaysSinceFipronil.csv) includes data from captures and combings of prairie dogs on the treated sites only, in 2020 and 2022 (when decent numbers of fleas were found on prairie dogs at the treated sites), and lists the following information for each prairie dog: YEAR (2020 or 2022), DAYSSINCE (days since fipronil grain treatment), and FLEAS (the number of fleas combed from the prairie dog). These data were used to evaluate rebounding flea numbers on the treated sites over time (under the question, how long does it take for flea numbers to rebound on sites treated with fipronil grain?). Day of sampling was not included in this analysis, because that variable would have been confounded with DAYSSINCE.