The end-Permian mass extinction (EPME; ca. 251.94 Ma) is the most severe mass extinction in the geological record. Detailed paleobiological investigations show a very rapid EPME event, and recently published delta U-238 data show a large negative excursion and thus a massive shift to globally expanded anoxia at the main extinction phase in the latest Permian. The negative shift in delta U-238 is in correlation with a globally characterized negative delta C-13 excursion near the Permian-Triassic boundary (PTB). In some highly expanded PTB carbonate sections, however, there are two distinct negative delta C-13 excursions whereas uranium isotopes (delta U-238) from such sections have not yet been examined, leaving a gap in the understanding of the global perturbations of marine redox conditions immediately following the EPME. Here, we present a new delta U-238 study of syndepositional dolostones from a well-characterized and highly expanded drill core, which recorded two pronounced negative delta C-13 excursions across the PTB, from the Carnic Alps, Austria. This drill core extends 331-meters across the PTB and provides a unique opportunity to explore the detailed timing, duration, and extent of marine redox chemistry changes before, during, and immediately after the EPME. Our new delta U-238 record shows two negative shifts, which are correlated with the two negative delta C-13 excursions. The first negative delta U-238 excursion preceding the EPME confirms the recently published delta U-238 records from across the EPME and support that syndepositional marine dolostones can record delta U-238 trends of seawater similar to that of limestones. Modeling of uranium isotope cycling in the latest Permian and earliest Triassic oceans suggests two distinct stages of expanded marine anoxia separated by a brief interval (similar to 100 kyr) of reoxygenation across the PTB. The first anoxic episode lasted for 60 kyr while anoxic seafloor area expanded to cover >18% of the entire seafloor, coeval with the main EPME horizon, agreeing with marine anoxia as a proximate kill mechanism for the EPME. The second anoxic event was less intense compared to the first anoxic pulse but sustained for a longer duration. A global modeling of coupled C, P, and U cycles show that two pulses of volcanic carbon injection that drives global warming and increased phosphorus weathering rate can reasonably reproduce our data to match two phases of anoxia. The model also demonstrates that the loss of terres-trial vegetation in the EPME is crucial to generating an intervening interval of oxygenated ocean. Our new study adds to a growing body of evidence that the global marine redox conditions underwent rapid oscillations during the EPME event and continued afterward, which may have played a central role in delaying the marine ecosystem recovery in the Early Triassic. (c) 2020 Elsevier Ltd. All rights reserved.