With the recent development of molecular markers, strategies for identifying patients with colorectal cancer (CRC) having a high risk of postoperative early relapse (within 1 y) and relapse have been improved. We previously constructed a multigene biochip with 19 candidate genes. The objective of the present study was to optimize a multigene biochip for detecting the risk of postoperative early relapse and relapse in patients with CRC. We included 357 patients with stage I-III CRC who underwent curative resection at a single institution between June 2010 and May 2015. During each follow-up, a postoperative surveillance strategy including the National Comprehensive Cancer Network recommendations and a multigene biochip was used. A statistical algorithm was developed to select candidate biomarkers for an optimal combination. After a 30.9-mo median follow-up, 67 patients (18.8%) had postoperative relapse, of whom 25 (7.0%) relapsed within 1 y after operation and accounted for 37.3% of all relapsed patients. Of the 19 circulating biomarkers, ELAVL4, PTTG1, BIRC5, PDE6D, CHRNB1, MMP13, and PSG2, which presented significant predictive validity, were selected for combination. The expression of the seven-biomarker biochip resulted in area under the receiver operating characteristic curve values of 0.854 (95% confidence interval: 0.756-0.952) for early relapse and 0.884 (95% confidence interval: 0.830-0.939) for relapse. Moreover, the sensitivity, specificity, and predictive accuracy levels were 84.0%, 83.1%, and 83.2% for early relapse and 76.1%, 91.0%, and 88.2% for relapse (P = 0.415, 0.006, and 0.054, respectively). The median lead times before the detection of postoperative early relapse and relapse were 3.8 and 10.4 mo, respectively. From 19 circulating biomarkers, we optimized seven contemporary circulating biomarkers. The prediction model used for the early and accurate identification of Taiwanese patients with CRC having a high risk of postoperative early relapse and relapse seems to be feasible and comparable.