Abstract：The diagnostic fingerprints of the needles of P. massoniana Lamb with different resin-yield capacity were analyzed by thermal desorption GC/MS and computer image processing. Results show that with the temperature 350℃ of thermal desorption, volatile constituents in the needles are well separated and their fingerprints are well repeated. There are 31 common constituents in the fingerprint of high-yield resin P. massoniana Lamb and in that of low-yield resin P. massoniana Lamb. Of which T values of 23 common constituents are greater in high-yield resin P. massoniana Lamb than what in low-yield resin P. massoniana Lamb. T values of 9 common constituents are over 50% and they concentrate on the segment where value is between 1.36 to 2.15. By analysis with fuzzy clustering, both similarity coefficient of high-yield resin P. massoniana Lamb and that of low-yield resin P. massoniana Lamb are over 0.90.