Article–commenting functionality allows users to add publicly visible comments to an article on a publisher’s website. As well as facilitating forms of post-publication peer review, for publishers of open-access mega-journals (large, broad scope, open-access journals that seek to publish all technically or scientifically sound research) comments are also thought to serve as a means for the community to discuss and communicate the significance and novelty of the research, factors which are not assessed during peer review. In this article we present the results of an analysis of commenting on articles published by the Public Library of Science (PLOS), publisher of the first and best-known mega-journal PLOS ONE, between 2003 and 2016. We find that while overall commenting rates are low, and have declined since 2010, there is substantial variation across different PLOS titles. Using a typology of comments developed for this research, we also find that only around half of comments engage in an academic discussion of the article and that these discussions are most likely to focus on the paper’s technical soundness. Our results suggest that publishers are yet to encourage significant numbers of readers to leave comments, with implications for the effectiveness of commenting as a means of collecting and communicating community perceptions of an article’s importance.