Early diagnosis is significant for the treatment of esophageal cancer. Surface enhanced Raman scattering(SERS) spectroscopy has been applied in this article to detect the spectra fo r hemoglobin from 24esophageal cancer patients and 20healthy persons,while silver colloid is used as the SERS sub strate.Principal component analysis (PCA) and discriminate analysis are employed to analyze the spectra of the two groups. From the SERS spectra in this experiment, it could be found that there are some differences between the spectra of the hem oglobin from the esophageal cancer patients and the healthy persons,which may be caused by the different concentrations of s ubstances in the hemoglobin.The results demonstrate that compared with the healthy persons,there are more iron ions a t low spin state in the hemoglobin of esophageal cancer patients.The three-dimensional scatter plot of PC scores got by statistical analysis shows a good discrimination for the health and cancer groups as they distribute in relatively separate areas.The diagnosis sensitivity is 96%,the diagnosis specificity is 85%,and the total accuracy of this method is 91%.The results show good potential for early diagnose of esophageal cancer by statistical analysis of SERS spectra of h emoglobin.