Comparative genomics has contributed to the growing evidence that sexual selection is an important component of evolutionary divergence and speciation. Divergence by sexual selection is implicated in faster rates of divergence of the X chromosome and of genes thought to underlie sexually selected traits, including genes that are sex biased in expression. However, accurately inferring the relative importance of complex and interacting forms of natural selection, demography, and neutral processes that occurred in the evolutionary past is challenging. Experimental evolution provides an opportunity to apply controlled treatments for multiple generations and examine the consequent genomic divergence. Here, we altered sexual selection intensity, elevating sexual selection in polyandrous lines and eliminating it in monogamous lines, and examined patterns of allele frequency divergence in the genome of Drosophila pseudoobscura after more than 160 generations of experimental evolution. Divergence is not uniform across the genome but concentrated in islands, many of which contain candidate genes implicated in mating behaviors and other sexually selected phenotypes. These are more often seen on the X chromosome, which also shows greater divergence in F-ST than neutral expectations. There are characteristic signatures of selection seen in these regions, with lower diversity on the X chromosome than the autosomes, and differences in diversity on the autosomes between selection regimes. Reduced Tajima's D within some of the divergent regions may imply that selective sweeps have occurred, despite considerable recombination. These changes are associated with both differential gene expression between the lines and sex-biased gene expression within the lines. Our results are very similar to those thought to implicate sexual selection in divergence between species and natural populations, and hence provide experimental support for the likely role of sexual selection in driving such types of genetic divergence, but also illustrate how variable outcomes can be for different genomic regions.