Late Eocene sandstone capping hills and small buttes in the Medicine Pole Hills in southwestern North Dakota has been correlated with the Chalky Buttes Member of the Chadron Formation. Heavy-mineral analysis was used to assess this correlation. Optical microscopy and chemical analysis using a scanning electron microscope/energy dispersive spectrometer (SEM/EDS) system were used to determine abundances and compositions of heavy minerals separated from 0.25–0.30 mm fractions of nine samples from seven depositional units (353–602 grains per sample). The 0.063–0.125 mm fraction was also studied for one of these samples. Samples ranged from coarse- to medium- to fine-grained poorly consolidated sandstone. Total heavy mineral recovery (0.4–14.1%) generally corresponded with the grain-size distribution of samples, with higher recoveries in coarser samples. Opaque grains (ferruginous aggregates, Fe-Ti oxides, and leucoxene) comprised 0.8–15.6 percent of heavy mineral grains. Non-opaque grains were dominated by amphibole (24.4–52.2%) and epidote-clinozoisite (10.6–38.1%), with variable diopside (0–51.1%). Generally less abundant minerals included garnet, apatite (bone fragments), biotite, and muscovite. One sample was biotite-rich (36.5%). Chemical compositions of heavy minerals showed no significant stratigraphic variations, suggesting the nature of the sources did not vary. Diopside, biotite, and approximately 1/3 of the amphibole grains were interpreted as derived from Tertiary volcanic sources. The majority of the amphibole grains (tremolite and blue-green hornblende), epidote, and garnet were derived from Precambrian metamorphic sources. The relative contribution of volcanic sources was variable (19–64%), being highest in those samples with more diopside and (with the exception of the biotite-rich sample) highest in those samples with higher overall heavy-mineral abundances. Individual heavy-mineral abundances also correlated fairly well with grain size. Overall, variations in heavy minerals were explained by variable volcanic source contributions and variations in grain-size distributions of the samples. Despite the variability found, cluster analysis showed that the samples from Medicine Pole Hills were consistently different than the generally much more stable heavy-mineral assemblages found in the Chalky Buttes Member. They were also consistently different when volcanic heavy minerals were ignored. The differences could not be explained by variable volcanic contributions, grain-size effects, and/or differences in heavy mineral stabilities alone. They require differences in provenance, suggesting that the sandstone of Medicine Pole Hills should not be correlated with the Chalky Buttes Member, or, at the very least, it should be considered a distinct facies of the Chalky Buttes Member.