Antarctic krill and salps are major macroplankton contributors to Southern Ocean food webs and biogeochemical cycles. Sustainably managing the fishery for Antarctic krill against a backdrop of rapid regional climate change requires information on distribution and changes over time. Many data have been collected on the abundance of the two taxa from net sampling surveys since 1926, but they are stored in national archives, sometimes only in archived notebooks. An international team of scientists has collated these disparate data into a central database called KRILLBASE, a composite, circumpolar database of numerical density (no.m-2) of postlarval Euphausia superba and salps (aggregate and solitaries and species combined), with associated environmental information, metadata and documentation. The aim is to provide a temporal-spatial data resource to support studies of Southern Ocean ecology, including studies in support of fisheries management and conservation. Previous versions of KRILLBASE have led to a series of papers since 2004 that illustrate some of its potential uses. The latest version contains data on krill from ~12,520 hauls and data on salps from ~10,832 hauls. These span 49 seasons in two epochs (1926-1939 and 1976-2011), including over 215 individual surveys from 11 nations, and data from both monitoring programs and large synoptic surveys. Due to variations in sampling method KRILLBASE includes additional standardised krill density data representing the expected density from a relatively efficient sampling method. This document illustrates the coverage and variability in sampling, and provides circumpolar distribution maps of krill and salps, as well as regional maps for the well-sampled Atlantic and Indian Ocean sectors of the Southern Ocean. The data were made available via a public website (https://www.bas.ac.uk/project/krillbase/) and the Digital Object Identifier http://doi.org/brg8 in October 2016. A forthcoming data paper will provide information on how to screen, interpret and use KRILLBASE to reduce artefacts in interpretation.