The structure of the human leukocyte-common antigen-related molecule (LAR) protein tyrosine phosphatase gene was elucidated using phage and cosmid genomic DNA clones. The LAR gene is composed of 33 exons spanning over 85 kilobase pairs. Exon 2 encodes the signal sequence and the first four amino acids in the mature LAR protein. The three immunoglobulin-like domains are encoded by exons 3-7, and the eight fibronectin type III (Fn-III) domains by exons 8-17. Exons 18-22 encode the juxta-membrane and transmembrane domains, and exons 23-33 encode the two conserved tyrosine phosphatase domains and the entire 3'-untranslated region. Exon 1, which presumably encodes the 5'-untranslated sequence, has not been identified. Reverse transcription-polymerase chain reaction analysis revealed the alternative splicing of a mini-exon (exon 13) in the Fn-III domain 5 of human LAR and other related genes (rat LAR, rat PTP sigma, and human PTP delta). RNase protection analysis showed that the human LAR mRNA in which exon 13 is spliced-out is the major mRNA species in all cell lines examined. Reverse transcription-polymerase chain reaction analysis revealed further alternative splicing of LAR mRNA involving the Fn-III domains 4, 5, 6, and 7 in various combinations. These findings will facilitate the understanding of the physiological functions of the LAR extracellular domain.