Background: In pts with CRC, the clinical usefulness of ctDNA testing after curative resection has been strongly suggested. Prognostic biomarkers are also needed for pts with resectable metastases to identify high-risk patients who may be spared unnecessary surgery or local ablation. We aimed to create a prediction model to estimate the risk of post-operative recurrence using an integrated analysis of preoperative ctDNA level and radiographic imaging. Methods: The GALAXY study (UMIN000039205) included 645 pts with newly diagnosed or recurrent CRC with resectable metastatic lesions. Pts without liver or lung metastases and those who had received chemotherapy within 3 months prior to ctDNA measurement were excluded. We used the ratio of preoperative ctDNA levels [mean tumor molecules (MTM)/mL] to tumor metastasis volume (mL) from radiographic imaging (ctDNA/volume) to predict recurrence. Cutoff ctDNA/volume values were determined from receiver operating characteristic curves using Youden index. Pts with above and below the cutoff values were defined as high and low risk, respectively. Disease-free survival (DFS) was calculated from the date of definitive surgical resection to first recurrence or death. Results: In this cohort of 110 pts with liver metastases and 24 with lung metastases, the median follow-up was 13.2 months (range: 0.7-30.9). The cutoff values for the ctDNA/volume model were 0.837 (liver) and 0.496 (lung) MTM/mL². Among the pts with liver metastases, the high-risk group predicted by the ctDNA/volume model had a median DFS of 14.7 months (95% CI: 6.1–not reached) versus not reached (95% CI: not reached–not reached) in the low-risk group (HR = 3.6, P = 0.021). Among the pts with lung metastases, the median DFS was 2.9 months (95% CI: 2.1–not reached) in the high-risk group versus 21.8 months (95% CI: 21.8–not reached) in the low-risk group (HR = 21.0, P< 0.0001). The high-risk group had a significantly higher cumulative recurrence at 6 months (17% and 67% in pts with liver and lung metastases, respectively) versus the low-risk group (0%, P = 0.012 and 6%, P< 0.001 in pts with liver and lung metastases, respectively) considering death as a competing risk. Conclusions: The ctDNA/volume model was able to predict postoperative recurrence in CRC pts with liver and lung metastases. Ongoing and future studies will further optimize the model by incorporating other clinicopathologic factors associated with postoperative CRC recurrence.