Farmer participation in agricultural research is increasingly seen as a powerful methodology to increase the relevance of technologies developed to benefit farmers' communities. In plant breeding, farmer participation is considered as a way to increase the probability of adoption of new varieties. However, the higher expected cost of participatory plant breeding (PPB) is seen as one of the main obstacles to its wider adoption.This paper addresses the issue of the different costs to an Institution of running a PPB program or a nonparticipatory program and uses the barley-breeding program at the International Center for Agricultural Research in the Dry Areas (ICARDA) as a case study.Observations and data collection were carried out during one full cropping season on the cost of the D. Mangione Food and Agriculture Organization of the United Nations three main components of the breeding program, i.e. the management of the field trials (land and seedbed preparation, planting, fertilizer application, weed control, harvesting, and seed threshing, cleaning, treating and packaging), the travel to farmers' fields or to the research sites, and the human resources (scientists, technical staff, local workers and farmers) involved in breeding activities.We compared two options for the centralized-nonparticipatory breeding program, differing in the number of sites (8 and 16) used for the on-farm trials, with 160 options for the decentralized-participatory breeding program, differing in the combination of number of sites (from 4 to 16) and number of trials per site (from 1 to 10).The results show that in both decentralizedparticipatory and centralized-non-participatory plant breeding the cost of managing the field trials is the highest followed by the cost of human resources and that of travel: the contribution of each component to the total cost varies with the various options and the various combinations of the number of sites and of farmers.The comparison of the aggregated costs indicates that in the case of the ICARDA' barley-breeding program there are no relevant differences between the participatory and the non-participatory plant breeding programs. This is largely associated with the fact that the decentralized-participatory breeding program reaches the same level of development of the breeding material 3 years earlier than the centralizednon-participatory breeding program. Depending on Springer 290 Euphytica (2006) 150:289-306 the type of centralized-breeding program and on the combination of number of sites and number of farmers per site in the participatory program, the aggregated costs of the participatory program are lower than those of the centralized-breeding program by between 5 and 28%. At the same level of cost of the centralized program, the model of participatory program used in this study generates more information due to the use of more trials at each site. This improves selection efficiency and provides an analytical tool to optimize the number of sites and of farmers per site.