We investigated the possibility of positive selection acting on members of the putative seven-pass chemoreceptor superfamily in Caenorhabditis elegans, which comprises Ϸ1,300 genes encoding seven-pass G protein-coupled receptors (GPCRs). Using a maximum-likelihood approach, we conducted statistical tests for evidence of codon sites where the ratio of nonsynonymous substitutions per nonsynonymous site to synonymous substitutions per synonymous site (d N͞dS) was >1. Evidence for positive selection was found only for the srz family, about which virtually nothing specific is known. We extended the annotation of the srz gene family, establishing gene models for 60 srz genes in C. elegans and 28 srz genes in Caenorhabditis briggsae. d N͞dS ratios varied dramatically in different regions of the SRZ proteins, peaking in predicted extracellular regions. These regions included 23 sites where evidence of positive selection was highly significant, corresponding remarkably well with regions implicated in ligand binding in other GPCR family members. We interpret these results as indicating that the srz family is under positive selection, probably driven by ligand binding.positive selection ͉ ligand binding ͉ maximum likelihood ͉ synonymous ͉ nonsynonymous C aenorhabditis elegans has Ϸ1,300 predicted genes that encode members of putative chemosensory receptors and together define the seven-pass receptor (SR) superfamily (1-4), which belongs to the broader class of G protein-coupled receptors (GPCRs). Based on sequence alignment and phylogenetic analysis, SR superfamily members fall into about a dozen families. These families range in size from the large srh and str families (a few hundred genes each) to the modestly sized sra and srv families (Ϸ30 genes each). Each SR family appears to have arisen by gene duplication and divergence from a founder gene. These duplications have occurred sporadically over a long evolutionary period, giving rise to complex relationships. Near one extreme, str-5 and str-6 result from a recent duplication and differ by only two nucleotides in their coding regions. Near the other extreme, the str-1 and str-47 proteins are only 19% identical and presumably arose from an ancient duplication. Members of different SR families are even more distantly related, with the most distant pairs barely alignable.Proteins in the SR superfamily appear to be more rapidly diverging than the average gene (2, 3, 5, 6). For example, the average ortholog pair between C. elegans and Caenorhabditis briggsae has 80% amino acid identity (5), whereas the average str pair from the same data set has 59% identity (J.H.T., unpublished data). Rapidly diverging proteins may result from relaxed selective constraints, in which changes in protein sequence are relatively well tolerated. Alternatively, they may result from selective pressure for changes in amino acid sequence (positive selection or diversifying selection). In extreme cases, positive selection acts on all or most of a protein and can be detected in pairwise d N ͞d S ratio...