‘Breeding by numbers’ is the term commonly used to deride the practical application of quantitative genetics. However, even the most subjective methods of assessment of breeding stock actually involve breeding by numbers: whenever a judge has to choose a winner and a runner-up from a line of animals, he/she evaluates each animal for each of several traits, then assesses the relative importance of each trait, and then ranks the animals on a weighted sum of the traits, where the weights reflect relative importance. And if the aim is to rank animals in terms of how good their offspring will be, then the mental gymnastics also involves unwittingly taking account of all the relevant genetic and phenotypic parameters (heritabilities and correlations) of each of the traits and combinations of traits. This has been the situation for as long as animals have been domesticated. The only thing that is different with the quantitative-genetics approach is that the guessing and mental gymnastics inherent in the traditional method of assessment have been replaced by use of a transparent set of numbers that reflect actual performance (or subjective assessment of certain traits) that is directly relevant to profitability, followed by a set of calculations using the best-available estimates of genetic and phenotypic parameters, ending up with the best-available prediction of how the offspring of each animal will perform. Having entered the genomics age, humans will still be breeding animals by numbers in the decades to come, but the numbers will be informed by knowledge of the action of thousands of individual genes.