The invention discloses an establishing method for a lung cancer characteristic metabolite fingerprint spectrum in urine. The gas chromatography and mass spectrum combined technology is adopted to detect metabolic profiles of urine samples of lung cancer patients and healthy donors, and is combined with the biological method of a support vector machine to build a smoker lung cancer metabolite fingerprint spectrum and a non-smoker lung cancer metabolite fingerprint spectrum, wherein the smoker lung cancer metabolite fingerprint spectrum comprises oxalic acid, phosphoric acid, uracil, threonine, 5-pidolic acid, citric acid and galactose or comprises phosphoric acid, uric acid, citric acid and oxalic acid; and the non-smoker lung cancer metabolite fingerprint spectrum comprises alanine, oxalic acid, phosphoric acid, uracil, serine, threonine, 5-pidolic acid, ribose, aconitate, citric acid, galactose, tyrosine, palmitic acid and stearic acid. The method can detect only needing the urine samples, satisfies the requirements of effectiveness, simplicity and feasibility in the operation aspect, and is suitable for large-scale application.