PURPOSE: The variety of outcomes in patients with stage II colorectal carcinomas might be due to understaging caused by an inadequate number of lymph nodes (LNs) being examined. The aim of this study was to determine if any number of examined LNs reflects a reliable node-negative staging for colorectal carcinomas (CRCs). METHODS: Data on 241 patients (132 males) who underwent potentially curative resections for pT3 and pT4 CRC were reviewed. The patients ranged in age from 21 to 87 (mean: 58.2) years with a median follow-up of 43 (range: 7~96) months. The relationship between the number of LNs harvested and both the 5-year disease-free survival (DFS) and the overall survival (OS) rates were assessed for stage II CRCs. RESULTS: A median of 15 LNs (range: 3~104) was harvested per tumor specimen, and lymph-node metastases were present in 107 cases (44.4%). The proportion of lymph-node metastases increased as a function of the number of LNs harvested (P=0.0002; 95% confidence interval, 0.3333~0.8138). The number of LNs revealed to be the best number for dividing stage II patients into subgroups with different DFS and OS rates was ten. The 5-year DFS and OS rates of the 48 patients (35.8%) with nine or fewer LNs harvested were 68.6% and 76.8%, respectively, whereas those of the 86 patients (64.2%) with ten or more LNs harvested were 87.2% and 91.9%, respectively (DFS, P=0.0082; OS, P=0.0303). Moreover, there were no statistical differences between the node-negative patients with nine or fewer LNs harvested and the 67 stage III patients with N1 in respect to the DFS (68.6% vs. 56.7%; P= 0.2031) and the OS (76.8% vs. 68.3%; P=0.2772) rates. CONCLUSIONS: This study suggests that examining a greater number of lymph nodes increases the likelihood of accurate nodal staging and that a minimum of ten LNs per surgical specimen should be harvested and examined to label a pT3 or pT4 CRC as node-negative.