485 land race populations of barley collected from 91 Minimum Package areas spread over 12 provinces of Ethiopia were grown in the field and classified for number of kernel rows, spike length, awn type, spike density, rachilla hair, caryopsis, and kernel color. The different characters showed different levels of diversity by the Shannon-Weaver diversity index. Provinces were grouped into regions to detect possible differences in diversity within and among regions. Much of the variation in diversity is due to differences among characters; and no differences were detected among provinces and regions in the level of diversity, which could be due to insufficient sampling. A high diversity index (H'=0.72) was obtained for the Southwestern Plateau, and this was not significantly different from the value for Ethiopia (H'=0.68). Clinal patterns of variations were observed for some characters and localized concentrations for others. An attempt was made to explain these patterns based on climatic data and selection. Vavilov's theory of the emancipation of recessives was found to be untenable. A center of origin of barley is defined on the Arussi-Bale Highland in accordance with the requirements of the current concept of the center of origin; and the inadequacy of this concept is discussed. Disruptive selection is considered as the most probable cause of the maintenance of the diverse forms of barley in Ethiopia; and further detailed studies of the patterns of variation of barley in Ethiopia are recommended.