This study aimed to identify novel molecular markers for the early detection of colorectal cancer liver metastasis. Genes related to hepatic metastasis were screened from the Oncomine database. The candidate markers were tested by immunohistochemistry, and their predictive accuracy was assessed by the cross-validation method and an independent test set. We got three datasets containing 2,973 genes that were highly expressed in primary colon cancer tissues compared with non-metastatic colon cancer tissues and identified 7 candidate molecules for immunohistochemical validation. A total of 213 colorectal cancer samples were randomly divided into a training set (113 cases) and a blind testing set (100 cases). In the training set, immunohistochemical analysis showed that HP, OPN, and PTGIS expression were significantly higher in the hepatic metastasis group than in the non-metastasis group. Logistic regression analysis showed that HP and OPN levels in primary tumors and lymph node metastasis status were the only significant (P < 0.05) parameters for detecting liver metastasis. The predictive accuracy of markers was assessed by the cross-validation method and an independent test set. In leave-one-out cross-validation, the two markers combined with clinicopathologic features resulted in 91.2% sensitivity and 96.4% specificity for hepatic metastasis detection. In an independent test set, the combination achieved 94.5% sensitivity and 88.9% specificity for predicting the hepatic metastasis of colorectal cancer. Our results suggest that combined HP and OPN expression levels are significantly related to liver metastasis and prognosis, and, if this is validated, they could be used as accurate predictors of liver metastasis in colorectal cancer.