Silent sites in mammals have classically been assumed to be free front selective pressures. Consequently, the synonymous substitution rate (K-s) is often used as it proxy for the Mutation rate. Although accumulating evidence demonstrates that the assumption is not valid, the mechanism by which selection acts remain Unclear. Recent work has revealed that the presence of exonic splicing enhancers (ESEs) in coding sequence might influence synonyomous evolution. ESEs are predominantly located near intron-exon junctions, which may explain the reduced single-nucleotide polymorphism (SNP) density in these regions. Here we show that synonymous Sites in putative ESEs evolve more slowly than the remaining exonic sequence. Differential mutabilities of ESEs do not appear to explain this difference. We observe that Substitution frequency ill four-fold synonymous sites decreases its one approaches the ends of exons, consistent with the existing SNP data. This gradient is at least in part explained by ESEs being more abundant near junctions. Between-gene variation in K-s is hence partly explained by the proportion of the gene that acts as,in ESE. Given the relative abundance of ESEs and the reduced rates of synonymous divergence within them, we estimate that constraints on synonymous evolution within ESEs Causes the true mutation rate to be Underestimated by not more than similar to 8%. We also find that K-s Outside of ESEs is much lower in alternatively spliced exons than in constitutive exons, implying that other causes of selection on synonymous mutations exist. Additionally, selection on ESEs appears to affect nonsynonymous sites and may explain why amino acid usage near intron-exon junctions is nonrandom.