The present invention discloses a detection and analysis method for urine-modified nucleoside based on a surface-enhanced resonance Raman spectroscopy technology. In the method, a tumor marker modified nucleoside in the urine of a normal person and a cancer patient is extracted through the specificity of borophenylic acid gel; Au colloid is taken as an enhancing substrate to detect SERRS signals; and statistical analysis is conducted by using PLS-DA algorithms to establish a diagnosis and identification model for the SERRS of the urine-modified nucleoside. The model is used to discriminate that the to-be-detected urine-modified nucleoside belongs to the normal person or the cancer patient. After PLS-DA, the surface-enhanced resonance Raman spectroscopy data of the urine-modified nucleoside of the present invention has specificity of 96.9%, sensitivity of 98.2% and accuracy of 97.6%. The present invention has the characteristic of rapid and objective detection, and can provide important reference for doctors to diagnose esophagus cancer.