We reviewed 272 bird atlases (standardised surveys intended to document the distribution of birds) from around the world. Atlases we located were conducted in 50 countries from six continents with most (82.4%) from Europe and North America. Atlases were mostly run by ornithological societies (67.1%), had amassed at least 27.9 million records of birds over an area roughly 31.4% of the land area of the Earth, and had involved at least 108 000 contributors. They had a modal data collection period of 4 years (some ran over several decades) and varied greatly in scale, covering local areas to entire continents (21 km2 – 10 390 000 km2); atlases that covered larger areas involved more observers and generated more records. Most atlases (88.3%) were constrained to particular seasons, and most of these focussed on the main local breeding period (81.0%). Spatial sampling units ranged from 0.02 km2 (2 ha) to 3092 km2 and temporal units of sampling varied from 20 minutes to several years. Little information is available on the application of data generated by atlases. We focussed on five major atlases for which information was available. We located 97 scientific publications drawing on data from these five major atlases; papers most frequently focussed on bird distribution (26.8%), ecology (20.6%) and land-use planning (17.5%). Atlas books were cited often, 7–31 times per year. Provision of data to third parties from two major atlases (one from Australia and one from Britain and Ireland) was frequent and remarkably similar. Data were requested mostly for environmental impact studies (almost half of all requests), conservation policy and planning (~20%), research (~20%) and other mapping (~13%). Despite the uses we describe, atlas data seem under-utilised.