UNLABELLED: Purpose/aim: The appropriate staging of colorectal cancer requires at least 12 lymph nodes to be sampled. We evaluated whether lymph node sampling (LNS) and lymph node ratio (LNR) can predict the prognosis of stage II-III patients. MATERIALS AND METHODS: This is a retrospective study on 432 patients classified in LNS ≥12 and LNS <12. Disease-free survival (DFS) was computed using the Kaplan-Meier method. We stratified stage III patients into 4 quartiles base on LNR values. To determine the optimal LNR cut-off, receiver operating characteristic (ROC) curve analysis was performed. RESULTS: There was a positive association between the number of lymph node sampled and the number of metastatic lymph nodes (p < 0.01). Among stage II patients, the DFS was 81% for LNS ≥ 12 and 72% for LNS < 12 (p = 0.158). Among stage III patients, the DFS was 58% (p < 0.001). We found a significant association between LNR quartiles and relapse in stage III patients but only in the LNS ≥ 12 group. ROC curve analysis indicated an ideal LNR cut-off value at 0.194 (sensitivity 65% and specificity 61%). The DFS of patients with LNR below 0.194 was 71%, and that of patients with LNR above 0.194 was 45% (log-rank test, p < 0.001). In the patients with LNS ≥ 12, the cut-off of 0.257 could predict recurrence (specificity 86%). CONCLUSIONS: Stage II patients with LNS < 12 tend to have shorter DFS than stage II patients with LNS ≥ 12. In stage III patients, an appropriate LNR cut-off is a better prognostic predictor than LNR quartile, especially in patients with LNS ≥ 12.