The aim of this study was to find new protein biomarkers that could be used to detect hepatocellular carcinoma (HCC) in the serum. We identified 11 proteins in the tissue that could be used to classify samples from HCC and control subjects. The 11 identified tissue biomarkers were combined with 10 commonly used serum HCC biomarkers for further verification in a large number of serum samples from HCC patients and healthy controls. 17 of the 21 prospective serum biomarkers were determined to be differentially expressed through collinearity and significance analysis. Through the method of supervised learning, a random forest model was constructed to reduce the dimensionality of the number of differentially expressed proteins, and finally, 4 differentially expressed proteins were identified: AFP, GDF15, CEACAM-1, and MMP-9, and suggested to have potential application in clinical diagnosis of HCC.