cDNA encoding human 4-aminobutyrate aminotransferase (aminobutyrate:2-oxoglutarate aminotransferase) was prepared by polymerase chain reaction using mRNA from human neuroblastoma cells as the template and oligonucleotides synthesized on the basis of the information obtained from direct protein sequencing. The cDNA-deduced sequence enabled peptides, sequenced by automated Edman degradation, to be aligned for confirmation of the complete primary structure. The results are compared with the recently published sequence of the rat enzyme deduced entirely from DNA sequencing [Medina-Kauwe, L. K., Tillakaratne, N. J. K., Wu, J.-Y. & Tobin, A. J. (1994) J. Neurochem. 62, 1267-1275]. Although the sequences are almost identical for most of their length, they differ in a segment of 36 residues. Almost complete identity of the two sequences is established if it is assumed that a frame-shift error was introduced into the reported rat cDNA sequence. The human cDNA was used to probe for the presence of 4-aminobutyrate aminotransferase mRNA in human tissues and a significant transcript was found in heart, placenta and in tissues usually associated with the expression of this enzyme.