Gene duplication is one of the primary forces of evolutionary change. We present data from three different pericentromeric regions of human chromosomes, which indicate that such regions of the genome have been sites of recent genomic duplication. This form of duplication has involved the evolutionary movement of segments of genomic material, including both intronic and exonic sequence, from diverse regions of the genome toward the pericentromeric regions. Sequence analyses of the target sites of duplication have identified a novel class of interspersed GC-rich repeats located precisely at the boundaries of duplication. Estimates of the evolutionary age of these duplications indicate that they have occurred between 10 and 25 mya. In contrast, comparative analyses confirm that the GC-rich pericentromeric repeats have existed within the pericentromeric regions of primate chromosomes before the divergence of the cercopithecoid and hominoid lineages (∼30 mya). These data provide molecular evidence for considerable interchromosomal duplication of genic segments during the evolution of the hominoid genome and strongly implicate GC-rich repeat elements as playing a direct role in the pericentromeric localization of these events Genome evolution is dependent on the processes of single-base-pair mutation and gene duplication. Duplication of a gene followed by mutation is responsible for the emergence of new genes with specialized functions in an evolving species. Two distinct molecular mechanisms of gene duplication are generally recognized. Tandem duplication of an ancestral gene, vis-à-vis processes of unequal crossover, produces a clustered family of related genes (Smith 1976). In contrast, whole-genome duplication (polyploidy) followed by chromosomal rearrangement and the re-establishment of the disomic state has been proposed as a mechanism for the interchromosomal duplication of large segments of a genome (Ohno 1970). The latter model, put forward originally by Susumu Ohno, explains observations of conserved genic synteny among nonhomologous chromosomes. Because of the genetic consequences of polyploidy, such events, although important in the early expansion of eukaryotic genomes, are rare and ancient (Ohno 1993;Wolfe and Shields 1997). Within the vertebrate lineage, for example, the last tetraploidization event leading to genome duplication is estimated to have occurred >400 mya (Lundin 1993). If polyploidy were the only model by which entire genes could be duplicated among nonhomologous chromosomes, it would then follow that interchromosomal paralogs of recent origin would be unexpected.Recent analyses of data from the Human Genome Project suggest a third form of genomic duplication, which is mechanistically distinct from polyploidy and tandem duplication models of genome evolution. Several independent reports indicate that genic segments, ranging in length from 5 kb to 30 kb, possess duplicate copies within the pericentromeric regions of human autosomes (Borden et