Construction of large collagen cDNA has been hindered by the relatively large size and high G-C content of processed mRNA. We describe here the development of a rapid and efficient method for obtaining large full-length collagen cDNA. A full-length (4.3 kb) murine pro alpha 2(I) collagen cDNA was constructed by synthesis of a first-strand cDNA library with use of poly-A RNA (MC3T3-E1) and the oligo-dT17-adapter primer described by Frohman et al (Proc Natl Acad Sci USA 85:8998, 1988). Pro alpha 2(I) collagen cDNA were specifically amplified by the polymerase chain reaction (PCR) with a pro alpha 2(I) specific primer as the 5' primer (20mer; corresponding to nucleotide positions 42-61 in the first exon of the murine pro alpha 2(I) collagen gene, COL1A2), and with the adapter sequence 5' to the dT17 as the 3' primer. The PCR conditions were optimized to allow amplification of the expected 4.0-5.0-kb product; a major 4.3-kb product was visualized by ethidium bromide, identified by in situ gel hybridization, and cloned. DNA sequencing determined that it contained the correct 5' sequence and the 3' end had a 68 basepair (bp) 3' untranslated region. The entire sequence that codes the amino-terminal propeptide domain has been determined and compared to the human sequence. The homology between human and mouse is less in the amino terminal propeptide than in the triple helical domain; exon 5 of murine COL1A2 codes for an additional six amino acids not found in human COL1A2.