The fitness costs of disease resistance in plants are reviewed and their relevance to plant breeding is discussed. An important theme of current research is that plant defence is closely integrated into pathways that regulate growth and development in plants, so mutations in genes with central locations in defence networks often have numerous pleiotropic effects. There is increasing evidence that resistance to one disease involves trade-offs with responses to other bio-antagonists; the numerous pleiotropic effects of mlo resistance to powdery mildew of barley are discussed as an example. There is a striking contrast between the low costs of resistance in most crop diseases and the sometimes high costs of resistance in model systems studied in fundamental research. It is suggested that, through natural and artificial selection over thousands of years, farmers and more recently plant breeders have selected alleles which maximize the benefits and minimize the costs of resistance. There is sometimes substantial genotype-by-environment interaction in fitness costs, which makes experiments on fitness in plants especially challenging, but it is essential to understand these interactions to appreciate the relevance of fitness costs and trade-offs to crops in field conditions. It is proposed that in the great majority of cases, in which it is not feasible to investigate fitness costs and their interactions with the environment in depth, plant breeding is an efficient way of incorporating genes with benefits which exceed any costs into commercially successful cultivars of arable crops.