Genomic selection is revolutionizing dairy cattle breeding by using genomic breeding values (GEBV) to make selection decisions. GEBV are calculated as the sum of effects of dense genetic markers or haplotypes across the entire genome, potentially capturing all quantitative trait loci (QTL) contributing to trait variation. The reliability of GEBV has been evaluated in experiments in the United States, New Zealand, Australia, and the Netherlands, with reliabilities ranging from 20% to 67% for young bulls without progeny test results. The reliability depends on heritability, the number of bulls in the reference population, and the statistical methods used. The BLUP method, which assumes a normal distribution of marker effects, performs similarly to more complex Bayesian methods. All countries included a polygenic effect in their GEBV calculations to capture genetic variance not associated with markers and to select for low-frequency QTL. The increase in reliability is significant, leading to at least two dairy breeding companies marketing bull teams based on GEBV alone at 2 years of age, potentially doubling the rate of genetic gain. Challenges include improving GEBV accuracy, integrating genomic information into national and international genetic evaluations, managing long-term genetic gain, and addressing computational challenges. Genomic selection can also be applied across breeds and to nonadditive effects, but requires careful consideration of LD and QTL effects.Genomic selection is revolutionizing dairy cattle breeding by using genomic breeding values (GEBV) to make selection decisions. GEBV are calculated as the sum of effects of dense genetic markers or haplotypes across the entire genome, potentially capturing all quantitative trait loci (QTL) contributing to trait variation. The reliability of GEBV has been evaluated in experiments in the United States, New Zealand, Australia, and the Netherlands, with reliabilities ranging from 20% to 67% for young bulls without progeny test results. The reliability depends on heritability, the number of bulls in the reference population, and the statistical methods used. The BLUP method, which assumes a normal distribution of marker effects, performs similarly to more complex Bayesian methods. All countries included a polygenic effect in their GEBV calculations to capture genetic variance not associated with markers and to select for low-frequency QTL. The increase in reliability is significant, leading to at least two dairy breeding companies marketing bull teams based on GEBV alone at 2 years of age, potentially doubling the rate of genetic gain. Challenges include improving GEBV accuracy, integrating genomic information into national and international genetic evaluations, managing long-term genetic gain, and addressing computational challenges. Genomic selection can also be applied across breeds and to nonadditive effects, but requires careful consideration of LD and QTL effects.