Publishing in top journals is crucial to academic careers, but many fields show substantial ethnic disparities in publication counts, which may be caused by disparities in producing manuscripts or getting them accepted. Here, we investigate the latter mechanism using the peer review files of 16.5K manuscripts submitted between 2013-2018 to a field-leading biology journal (Journal A) and a middle-tier journal of similar scope (Journal B). The editorial data are supplemented with authors’ name-inferred ethnicities and extensive controls including submissions’ topic, author prestige, and citation impact, even for rejected-and-published-elsewhere submissions. We find substantial disparities in acceptance across inferred ethnicities and that these are driven by the editors and not peer reviewers. In particular, for a given amount of future impact and other submission characteristics, papers by East Asian-named authors were 4.3-14.6 percentage points less likely to be accepted than those by British-origin-named authors. Journal A editors were about 7.1-8.1 percentage points less likely to send East Asian-authored (Chinese and non-Chinese) papers out for peer review and, for a given level of reviewer enthusiasm, 7.2 percentage points less likely to ultimately accept them (non-Chinese only). In contrast, peer reviewers gave recommendations that were similar across all name-inferred ethnicities. As science continues to globalize, these findings signal the need to better understand ethnic disparities in the review process.