Investigation of surface-enhanced Raman spectra of oxyhemoglobin for 30 liver cancers and 30 normal persons based on the silver nanofilm was reported. Principal components analysis (PCA) and independent variable T test were employed to develop effective diagnostic algorithms for distinguishing liver cancer form normal. Discriminant analysis was used to evaluate the diagnostic efficiency and the result indicates that SERS spectra of the oxyhemoglobin are obviously different between normal persons and liver cancers. PCA and independent variable T test were employed to get a three-dimensional scatter plot of PC scores for the healthy and cancer groups, and it can be learned that they are distributed in separate areas. By using the method of discriminate analysis, it was fount that the diagnostic algorithm separates the two groups with sensitivity of 96.7% and diagnostic specificity of 90%, the overall diagnostic accuracy was 93.3%. By analyzing the assignations of the SERS bands, it was found that the content of asparagine, tyrosine and phenylalanine in the hemoglobin are significantly lower than healthy people. The results from this exploratory study demonstrate that SERS detection of oxyhemoglobin for liver cancers based on the method of PCA combined with independent variable T test is expected to develop into a new type of liver cancer diagnostic tool.