The structure of the rat parvalbumin gene has been elucidated from analysis of six overlapping clones isolated from a rat lambda Charon 4A genomic library. Two of the clones were mapped in detail, and all exons were localized by Southern hybridization using fragments of a full-length parvalbumin cDNA (Epstein, P., Means, A. R., and Berchtold, M. W. (1986) J. Biol. Chem. 261, 5886-5891). The rat parvalbumin transcription unit is 15.5 kilobase pairs in length and contains four introns. The first intron divides the 5'-nontranslated region, whereas the other three interrupt coding DNA. All intron/extron boundaries were sequenced as was 377 base pairs immediately 5' from the putative transcription initiation site. The promoter region contains eukaryotic regulatory homologies to the "TATA" box at -24 and "CAAT" box at -47 and -156. In addition, two doublets consisting of 11-base pair direct repeats exist in the promoter region. Parvalbumin binds two Ca2+, whereas many other members of the same superfamily bind four. Comparison of the genes that encode these proteins provides a strong confirmation of the hypothesis that parvalbumin evolved from an ancestral gene specifying a four-domain Ca2+-binding protein. The rat parvalbumin gene was also utilized to assign its human counterpart to chromosome 22 from data obtained by hybridization to DNA from a somatic cell hybrid panel. It was also used to isolate a 7.5-kilobase pair EcoRI fragment from a human chromosome 22 DNA library.