The polyhomeotic (ph) locus in Drosophila melanogaster consists of the two tandemly duplicated genes ph-d (distal) and ph-p (proximal). They code for transcriptional repressors belonging to the Polycomb group proteins, which regulate homeotic genes and hundreds of other loci. Although the duplication of ph occurred at least 25 million to 30 million years ago, both copies are very similar to each other at both the DNA and the protein levels, probably because of the action of frequent gene conversion. Despite this homogenizing force, differential regulation of both transcriptional units suggests that the functions of the duplicates have begun to diverge. Here, we provide evidence that this functional divergence is driven by positive selection. Based on resequencing of an Ϸ30-kb region around the ph locus in an African sample of D. melanogaster X chromosomes, we identified a selective sweep, estimated its age and the strength of selection, and mapped the target of selection to a narrow interval of the ph-p gene. This noncoding region contains a large intron with several regulatory elements that are absent in the ph-d duplicate. Our results suggest that neofunctionalization has been achieved in the Drosophila ph genes through the action of strong positive selection and the inactivation of gene conversion in part of the gene.gene duplication ͉ selective sweep G ene duplication is a major evolutionary mechanism for generating new genes and new functions, thus increasing organismal complexity. Since Ohno's (1) pioneering ideas, this topic has become a main stay of evolutionary biology research. Ohta (2) laid the groundwork by exploring the population genetic mechanisms leading to the maintenance of gene duplications and the evolution of multigene families. The early evolution and functional divergence of duplicated genes, on the other hand, remained somewhat obscure. Theoretical work suggests that both genetic drift and positive selection may play a role in the fixation and early evolution of duplicate genes (3-5). In particular, positive selection is thought to drive the fixation of a duplicate gene that has gained a new function through acquisition of a beneficial mutation, a process referred to as neofunctionalization (6). However, there is currently limited evidence for this suggestion. The reasons for this may be twofold. First, it is difficult to detect newly diverging gene copies and, at the same time, identify selection at one and/or the other copy at the population level. Second, recent mathematical modeling predicts that neofunctionalization is unlikely in the presence of gene conversion, unless selection is very strong (7).One possible method for detecting strong positive selection and localizing the target of selection is the search for selective sweeps. A selective sweep denotes a signature of variation in the genome that results from the recent fixation of a new, strongly selected beneficial mutation (8) or standing low-frequency variants (9). Such footprints last not much longer than 0.1 N e generations,...