We aimed to develop a clinical predictive model for predicting the overall survival (OS) in stage I–III CRC patients after radical resection with normal preoperative CEA. This study included 1082 consecutive patients. They were further divided into a training set (70%) and a validation set (30%). The selection of variables for the model was informed by the Akaike information criterion. After that, the clinical predictive model was constructed, evaluated, and validated. The net reclassification index (NRI) and integrated discrimination improvement (IDI) were employed to compare the models. Age, histologic type, pT stage, pN stage, carbohydrate antigen 242 (CA242), and carbohydrate antigen 125 (CA125) were selected to establish a clinical prediction model for OS. The concordance index (C-index) (0.748 for the training set and 0.702 for the validation set) indicated that the nomogram had good discrimination ability. The decision curve analysis highlighted that the model has superior efficiency in clinical decision-making. NRI and IDI showed that the established nomogram markedly outperformed the TNM stage. The new clinical prediction model was notably superior to the AJCC 8th TNM stage, and it can be used to accurately assess the OS of stage I–III CRC patients undergoing radical resection with normal preoperative CEA.