We present data on the occurrences of Permian brachiopod families comprising the genera of 54 families, plus corals and fusulinids, at 343 stations. Geographic coordinates, age range, sampling index, lithology where possible, and data sources are given for each station. Q-mode cluster analysis based on presence or absence of brachiopod families reveals three distinct groups of localities, separable by a modified type of linear discriminant function. Discriminant scores for each group, calculated for every station, give sensitive measures of diversity that partially overcome deficiencies due to incomplete collecting. Localities with group A characteristics are found to have low diversity. They occur mainly in Australia – New Zealand and Siberia, are conspicuously devoid of fusulinids and corals, and are often in near association with glacial deposits. Group C localities are characterized by having high diversity, are usually associated with fusulinids and corals, but never with glacial deposits, and occur mainly in a sinuous belt roughly parallel to modern 30°N latitude. Group B localities are intermediate in diversity, sometimes associated with corals and fusulinids, and occur particularly in northern North America, parts of Russia, and much of South America. Group C is interpreted as indicative of a tropical biome, Group B of a temperate biome, and Group A as a subpolar to polar biome. Detailed regional analysis is made, age by age. A crude overall geographic distribution, independent of age, is made by taking average discriminant scores for each 10 degree block of latitude and longitude, to partially overcome uneven distribution of stations. Maps showing the resulting global distribution of biomes (and inferred paleoclimate and paleolatitude) indicate that the Dietz–Holden Permian reconstruction, based on bathymetry and paleomagnetics, provides a more logical continental framework than the present distribution of continents. Due to high-diversity faunas, and steep diversity gradients across the Baikal geosuture, we propose that the Chinese segment lay further south than in the Dietz–Holden model. Previous work on Permian brachiopod diversity gradients, which showed that the present continental configuration fitted the data better than Permian reconstructions, was based on too few data, lack of age control, and a faulty procedure to correct for incomplete sampling. Preliminary comparison of paleomagnetic data with brachiopod-paleoclimate data indicates general agreement, although this needs further study.