The human I(1)-imidazoline receptor candidate gene, iras, has previously been cloned and mapped to locus 3p21.1-9 (also known as Nischarin; accession No. AC006208). By comparison to a database of expressed sequence tags (ESTs), three alternatively spliced transcripts have been deduced. A map of 21 exons was constructed for the medium-length transcript (IRAS-M) containing 5,232 base pairs (bp) and encoding 1,504 amino acids (aas). Introns 13B and 13C are inserted into the two alternative transcripts, forming IRAS-S and IRAS-L mRNA (short and long isoforms). Northern blots confirmed the existence of these mRNA isoforms. In most brain regions the order of mRNA abundance was IRAS-M > IRAS-L > IRAS-S mRNA. Although aas 1 through 510 are theoretically identical, truncated proteins could be derived from IRAS-S (2,678 bp transcript yields 515 aas) and IRAS-L (9,457 bp transcript yields 583 aas). Because exon-16 of the iras gene has been proposed to encode the functional domains of imidazoline and a-5 integrin binding, only IRAS-M is expected to possess I(1) receptor properties. Subtype-selective cDNA expression constructs were therefore generated and used to transfect CHO cells. High-affinity I(1) binding was endowed by IRAS-M and IRAS-L, but not by IRAS-S transfection.