An accurate and precise geomagnetic polarity time scale is crucial to the development of a chronologic framework in which to test paleoclimatic and paleoenvironmental interpretations of marine and terrestrial records of the Eocene–Oligocene transition (EOT). The magnetic polarity patterns of relatively continuous marine and terrestrial records of the EOT have been dated using both radio-isotopic techniques and astronomical tuning, both of which can achieve a precision approaching ±30 k.y. for much of the Paleogene. However, the age of magnetic reversals between chrons C12n and C16n.2n has proved difficult to calibrate, with discrepancies of up to 250 k.y. between radio-isotopically dated and astronomically tuned marine successions, rising to 600 k.y. for comparisons with the 206Pb/238U-dated terrestrial record of the White River Group in North America. In this study, we reevaluate the magnetic polarity pattern of the Flagstaff Rim and Toadstool Geologic Park records of the White River Group (C12n–C16n.2n). Our interpretation of the Flagstaff Rim polarity record differs significantly from earlier studies, identifying a previously unreported normal polarity zone correlated to C15n, which eliminates discrepancies between the WRG and the 206Pb/238U-dated marine record of the Rupelian Global Stratotype Section and Point in the Italian Umbria-Marche basin. However, residual discrepancies persist between U-Pb–dated and astronomically tuned records of the EOT even when stratigraphic and systematic uncertainties associated with each locality and dating method are taken into account, which suggests that the uncertainties associated with astronomically tuned records of the EOT may have been underestimated.