Barley is one of the founder crops of the Neolithic transition in West Asia. While recent advances in genomics have provided a rather detailed picture of barley domestication, there are contradictory views on how the domestication process affected genetic diversity. We set out to revisit this question by integrating public DNA sequencing data from ancient barley and wide collections of extant wild and domesticated accessions. Using two previously overlooked approaches – analyses of chloroplast genomes and genome‐wide proportions of private variants – we found that the barley cultivated six millennia ago was genetically unique and more diverse when compared to extant landraces and cultivars. Moreover, the chloroplast genomes revealed a link between the ancient barley, an obscure wild genotype from north‐eastern Libya, and a distinct population of barley cultivated in Ethiopia/Eritrea. Based on these results, we hypothesize past existence of a wider North African population that included both wild and cultivated types and suffered from genetic erosion in the past six millennia, likely due to a rapid desertification that ended the Holocene African humid period. Besides providing clues about the origin of Ethiopian landraces, the hypothesis explains the post‐domestication loss of diversity observed in barley. Analyses of additional samples will be necessary to resolve the history of African barley and its contribution to the extant cultivated gene pool.