The relationship between protein and regulatory sequence evolution is a central question in molecular evolution. It is currently not known to what extent changes in gene expression are coupled with the evolution of protein coding sequences, or whether these changes differ among orthologs (species homologs) and paralogs (duplicate genes). Here, we develop a method to measure the extent of functionally relevant cis-regulatory sequence change in homologous genes, and validate it using microarray data and experimentally verified regulatory elements in different eukaryotic species. By comparing the genomes of Caenorhabditis elegans and C. briggsae, we found that protein and regulatory evolution is weakly coupled in orthologs but not paralogs, suggesting that selective pressure on gene expression and protein evolution is quite similar and persists for a significant amount of time following speciation but not gene duplication. Additionally, duplicates of both species exhibit a dramatic acceleration of both regulatory and protein evolution compared to orthologs, suggesting increased directional selection and/or relaxed selection on both gene expression patterns and protein function in duplicate genes.[Supplemental material is available online at www.genome.org.]The relative importance of coding sequence change versus regulatory sequence change in evolution has vexed evolutionary geneticists for over 50 years. Given recent genomic analyses showing the conservation of many proteins among distantly related taxa, it has been proposed that regulatory changes play a key role in generating the great morphological diversity present in multicellular species. However, little is known about the evolution of gene regulation or its relationship to protein evolution. Do highly conserved genes also show conserved expression patterns? Or can gene expression evolve independently from protein function? The former pattern is expected if strong stabilizing selection acts on genes as integrated units in which protein sequence and expression pattern are not dissociable. At the same time it has been argued that "developmental systems drift" may result in reorganization of regulatory systems as long as general developmental patterns are preserved (True and Haag 2001). If so, a high turnover of gene regulatory elements may uncouple cis-elementmediated gene expression and protein evolution.Differences in gene expression between species (or between duplicate genes) may entail changes in gene expression levels under the same conditions at the same developmental times, as well as gene expression changes in spatial, temporal, and environmental dimensions. Hereafter, we refer to the former changes as changes in expression magnitude and the latter as changes in relative expression. First attempts to find a correlation between the evolution of relative expression and protein evolution in yeast (using only duplicates) yielded contradictory results; one study argued that expression differences are not correlated with protein evolution (Wagner 2000), wher...