Thomas Hunt Morgan and colleagues identified variation in gene copy number in Drosophila in the 1920s and 1930s and linked such variation to phenotypic differences [Bridges CB (1936) Science 83:210]. Yet the extent of variation in the number of chromosomes, chromosomal regions, or gene copies, and the importance of this variation within species, remain poorly understood. Here, we focus on copy-number variation in Drosophila melanogaster. We characterize copy-number polymorphism (CNP) across genomic regions, and we contrast patterns to infer the evolutionary processes acting on this variation. Copy-number variation in D. melanogaster is nonrandomly distributed, presumably because of a mutational bias produced by tandem repeats or other mechanisms. Comparisons of coding and noncoding CNPs, however, reveal a strong effect of purifying selection in the removal of structural variation from functionally constrained regions. Most patterns of CNP in D. melanogaster suggest that negative selection and mutational biases are the primary agents responsible for shaping structural variation.centrality ͉ copy-number variation ͉ deletion ͉ duplication ͉ gene expression C opy-number polymorphism (CNP) has a dramatic impact on phenotypic variation within species. In humans, copyvariable regions account for Ͼ15% of the total detected genetic variation in gene expression (1), and some genes contributing to disease are contained within known duplication and deletion polymorphisms (2). In addition to its role in generating trait variation within species, CNP represents the raw material for gene family expansion and gene duplication between species. This raw material has apparently had a major role in evolution because 30-65% of genes in sequenced eukaryotes have been duplicated (3). On a larger scale, differences in the number, orientation, and distribution of chromosome segments are the most distinguishing features characterizing divergence in genome architecture between species. As in the case of gene duplication, the population genetic processes regulating CNP (and other variation) within species drive these exceptional differences in genome architecture (4).Although there is ample incentive to uncover the properties and dynamics of CNP, other than in humans little is known about copy-number variation in natural populations. Open questions remain about how much CNP exists in species' genomes. The observation that two unrelated healthy individuals can differ from one another in copy number across their genome raises uncertainty about the existence of an archetypal number of copies for any particular gene. Related to issues of the extent of CNP are differences in the type of CNP that can be found. Namely, the frequency, degree of dominance, and size of CNPs are largely unknown, as are differences between duplication and deletion polymorphisms. Equally important are the locations, chromosomal properties, and DNA sequence composition of CNPs. Finally, of all of the major issues surrounding CNPs, our knowledge of the evolutionary impli...