The presence of metastatic lymph nodes (MLNs) is the most important prognostic factor for gastric carcinoma, with the number of MLNs thought to be predictive of the prognosis. However, there have been long-standing debates on how to classify node-positive patients into prognostic groups appropriately. Recent findings in patients with colon and esophageal cancer have suggested that MLN size, more than MLN number, is an important prognostic factor; but less is known about the impact of MLN size on the prognosis of patients with gastric carcinoma. We therefore assessed the prognostic impact of large MLNs, especially those>or=2 cm, in patients with gastric carcinoma. A total of 1190 patients who underwent curative resection for gastric carcinoma between 2001 and 2003 and had lymph node metastases were divided into two groups according to the size of the largest MLN:>or=2 cm (n=51) vs. <2 cm (n=1139). Clinicopathologic data, including tumor recurrence and survival, were reviewed retrospectively. The median follow-up for living patients was 47 months (range 30-80 months). Age, sex ratio, type of surgery, and histologic classification did not correlate with MLN size. The depth of invasion did correlate with MLN size (T1-2 vs. T3-4, p=0.045) but not with the number of MLNs (N stage, p=0.311). The two groups showed similar distribution of stage according to the UICC/AJCC TNM staging system. Disease-free survival (34% vs. 53%, p<0.001) and overall survival (40% vs. 63%, p=0.011) were significantly worse in the large MLN group. Univariate analysis with the log-rank test showed that MLN>or=2 cm, type of surgery, T stage, N stage, and histologic classification had a significant impact on overall survival. Multivariate analysis with the Cox proportional hazard model showed that MLN>or=2 cm was an independent prognostic factor (hazard ratio 1.76, p=0.006), along with T stage and N stage. MLN>or=2 cm is an independent additional predictor of poor prognosis in patients with node-positive gastric carcinoma.