Oceanic anoxia has been proposed as a proximate cause of the end-Permian mass extinction (EPME), but evaluation of this hypothesis is hampered by limited detailed knowledge of its timing and extent. The recent development of uranium isotopes (δ238U) in carbonates as a global-ocean redox proxy provides new insights into this problem. Three earlier δ238U studies of Tethyan sections inferred development of extensive oceanic anoxia at the EPME. However, recent work raises concerns that diagenetic alteration may influence the reliability of δ238U records in bulk carbonate sediments. Here, we evaluate this possibility through δ238U analysis of a Permian-Triassic carbonate atoll section from the Panthalassic Ocean (Kamura, Japan) and comparison with existing δ238U profiles from the Tethys Ocean. The Kamura section exhibits a large negative δ238U shift across the EPME horizon identical both in timing and magnitude to those in Tethyan sections, demonstrating beyond a reasonable doubt that the negative seawater δ238U shift at the EPME was a global event, and that it was recorded by shallow carbonate facies globally. The robustness of the U-isotope proxy is further shown by the fact that a common global signal at the EPME was preserved despite major differences in the burial histories of Panthalassic and Tethyan sections, the former having been tectonically subducted and heated to greenschist metamorphic grade, whereas the latter accumulated in stable cratonic settings and experienced milder burial effects. Finally, we use leaching experiments to demonstrate that, although small-scale δ238U heterogeneity is common in both modern and ancient carbonates, it probably does not significantly affect bulk-carbonate δ238U trends.