Natural and artificial selection, geographical segregation and genetic drift can result in differentiation of allelic frequencies of single nucleotide polymorphism (SNP) at many loci in the animal genome. For individuals whose ancestors originated from different populations, their genetic compositions exhibit multiple components correlated with the genotypes or allele frequencies of these breeds or populations. Therefore, by using an appropriate statistical method, one can estimate the genomic contribution of each breed (ancestor) to the genome of each individual animal, which is referred to as the genomic breed composition (GBC). This paper reviews the principles, statistical methods and steps for estimating GBC of individual animals using SNP genotype data. Based on a linear regression model and an admixture model respectively, the protocols were demonstrated by the breed characterization of 198 purported Akaushi cattle, which included selection of reference SNPs and reference individual animals, and computing GBC for animals to be evaluated. The reference populations consist of 36 574 cattle from five cattle breeds (Akaushi, Angus, Hereford, Holstein and Jersey), each genotyped on either a 40K or 50K SNP chip. Four common SNP panels scanned from commercial chips for estimating GBC of individual animals are optimally selected, thereby expanding the functionalities of the currently available commercial SNP chips. It remains to be explored in future studies as to how estimated GBC can be incorporated to improve the accuracies on genomic prediction in purebred animals and crossbreds as well.