The human muscle-specific phosphoglycerate mutase encoding gene (PGAM-M) has been cloned from a genomic cosmid library and sequenced. The sequence corresponding to the coding region was evaluated and revised by sequencing of the protein itself, fully confirming our results. The amino acid sequence of the M isozyme presented a 80.6% homology with the B isozyme (non-muscle-specific isozyme), a value higher than previously reported. The PGAM-M gene is composed of three exons, which consist of 454, 180 and 202 bp, respectively, and are separated by two introns of 103 bp and approx. 5.6 kb, respectively. Comparison of the structure of the human PGAM-M gene with that coding for human bisphosphoglycerate mutase, an erythroid-specific enzyme belonging to the same multifunctional enzyme family, revealed that the location of the second intron is similar in each gene and corresponds to a tertiary subdomain in the spatial structure of the protein. The transcription start point (tsp) in the PGAM-M gene was identified by both primer extension and S1 nuclease-protection experiments. A TATA-box-like element was observed 29 bp upstream from the tsp; the sequence ATTGG, inverse/complementary to CCAAT-box, was found 40 bp upstream from the supposed TATA box. No muscle-specific consensus sequences could be detected in the 5'-untranslated region. Only one polyadenylation AATAAA signal was observed in the short 3'-untranslated region (43 bp long). Finally, only one copy of this gene is present in the human genome instead of the several copies found for the PGAM-B gene, suggesting the possible evolutionary origin of the muscle subunit in a modified copy of the PGAM-B gene.