We have developed webKnossos, an open-source web-based tool for visualization, annotation, and sharing of large-scale 3D image datasets. webKnossos enables fast data browsing of 3D datasets, sized from gigabytes to petabytes, in the browser. In addition to visualization of multi-channel raw microscopy data, the software can display segmentations, prediction maps, and other overlays. Data can be easily and securely shared between collaborators, annotators, and reviewers via links. webKnossos supports multiple annotation tools including volume labeling, line-segment annotations (aka skeletons), and proof-reading tools. These annotation tools can be used for visualization, generation of training and evaluation data, and correction of automated segmentations. For visual inspection of segmented objects, webKnossos features automatic mesh generation. webKnossos is extensible and scriptable through APIs and supports standard file formats for interoperability. As a scientific resource, webknossos.org serves as a database for published image datasets including their annotations. webKnossos has been published in Nature Methods (Boergens et al. 2017).