<jats:title>Abstract</jats:title><jats:sec> <jats:title>Background</jats:title> <jats:p>In this study, we used computed tomography (CT)-based radiomics signatures to predict the mutation status of <jats:italic>KRAS</jats:italic> in patients with colorectal cancer (CRC) and to identify the phase of radiomics signature with the most robust and high performance from triphasic enhanced CT.</jats:p> </jats:sec><jats:sec> <jats:title>Methods</jats:title> <jats:p>This study involved 447 patients who underwent <jats:italic>KRAS</jats:italic> mutation testing and preoperative triphasic enhanced CT. They were categorized into training (<jats:italic>n</jats:italic> = 313) and validation cohorts (<jats:italic>n</jats:italic> = 134) in a 7:3 ratio. Radiomics features were extracted using triphasic enhanced CT imaging. The Boruta algorithm was used to retain the features closely associated with <jats:italic>KRAS</jats:italic> mutations. The Random Forest (RF) algorithm was used to develop radiomics, clinical, and combined clinical–radiomics models for <jats:italic>KRAS</jats:italic> mutations. The receiver operating characteristic curve, calibration curve, and decision curve were used to evaluate the predictive performance and clinical usefulness of each model.</jats:p> </jats:sec><jats:sec> <jats:title>Results</jats:title> <jats:p>Age, CEA level, and clinical T stage were independent predictors of <jats:italic>KRAS</jats:italic> mutation status. After rigorous feature screening, four arterial phase (AP), three venous phase (VP), and seven delayed phase (DP) radiomics features were retained as the final signatures for predicting <jats:italic>KRAS</jats:italic> mutations. The DP models showed superior predictive performance compared to AP or VP models. The clinical–radiomics fusion model showed excellent performance, with an AUC, sensitivity, and specificity of 0.772, 0.792, and 0.646 in the training cohort, and 0.755, 0.724, and 0.684 in the validation cohort, respectively. The decision curve showed that the clinical–radiomics fusion model had more clinical practicality than the single clinical or radiomics model in predicting <jats:italic>KRAS</jats:italic> mutation status.</jats:p> </jats:sec><jats:sec> <jats:title>Conclusion</jats:title> <jats:p>The clinical–radiomics fusion model, which combines the clinical and DP radiomics model, has the best predictive performance for predicting the mutation status of <jats:italic>KRAS</jats:italic> in CRC, and the constructed model has been effectively verified by an internal validation cohort.</jats:p> </jats:sec>