Drought is one of the major factors limiting crop production worldwide. Dry areas are a much less homogeneous population of target environments than areas with high and reliable rainfall. In this paper we argue that a decentralized participatory plant breeding programme can address the complexity of dry areas, characterized by high and repeatable genotype × locations and genotype × years within locations interactions, more efficiently and effectively than a centralized non-participatory plant breeding programme. This is because varieties can be tailored not only to the multitude of target environments typical of dry areas, but also to diverse clients needs. In addition, varieties can be delivered in a shorter time and with a higher probability of adoption. Decentralized participatory plant breeding also has beneficial effects on biodiversity because selection is for specific adaptation rather than for broad spatial adaptation. The paper gives examples of methodological aspects including the modes of farmer selection, the precision of the trials, the efficiency of selection, the response to selection, the role of the type of germplasm and the role of molecular breeding in a participatory breeding programme. The paper gives the example of drought-resistant barley lines identified through extensive field testing and selection in a decentralized participatory breeding programme, and concludes that this type of plant breeding may be better targeted, more relevant and more appropriate for poor farmers in marginal areas.