Background: Patients with colorectal cancer commonly experience disturbances in coagulation homeostasis. Activation of the coagulation system contributes to cancer-associated thrombosis as the second risk factor for death in cancer patients. This study intended to discover coagulation-related genes and construct a risk model for colorectal cancer patients' prognosis. Methods: Coagulation-related genes were identified by searching coagulation-related pathways in the Molecular Signatures Database. Transcriptomic data and clinical data were downloaded from the Cancer Genome Atlas and Gene Expression Omnibus datasets. Univariate Cox and backward stepwise regression were utilized to identify prognosis-related genes and construct a predictive risk model for the training cohort. Next, survival analysis determines the risk model's predictive power, correlation with clinicopathological characteristics, and nomogram. Additionally, we characterized the variances in immune cell infiltration, somatic mutations, immune checkpoint molecules, biological functions, and drug sensitivity between the high- and low-score patients. Result: Eight hundred forty-five genes were obtained by searching the theme term ''coagulation'' after de-duplication. After univariate regression analysis, 69 genes correlated with prognosis were obtained from the Cancer Genome Atlas dataset. A signature consisting of 17 coagulation-related genes was established through backward stepwise regression. The Kaplan-Meier curve indicated a worse prognosis for high-score patients. Time-dependent receiver operating characteristic curve analysis demonstrated high accuracy in predicting overall survival. Further, the results were validated by two independent datasets (GSE39582 and GSE17536). Combined with clinicopathological characteristics, the risk model was proven to be an independent prognostic factor to predict poor pathological status and worse prognosis. Furthermore, high-score patients had significantly higher stromal cell infiltration. Low-score patients were associated with high infiltration of resting memory CD4+ T cells, activated CD4+ T cells, and T follicular helper cells. The low-score patients exhibited increased expression of immune checkpoint genes, and this might be relevant to their better prognosis. High-score patients exhibited lower IC50 values of Paclitaxel, Rapamycin, Temozolomide, Cyclophosphamide, etc. The differential signaling pathways mainly involve the calcium signaling pathway and the neuroactive ligand-receptor interaction. Lastly, a nomogram was constructed and showed a good prediction. Conclusion: The prognostic signature of 17 coagulation-related genes had significant prognostic value for colorectal cancer patients. We expect to improve treatment modalities and benefit more patients through research on molecular features.