In gastric cancer, the significance of lymphatic invasion as a prognostic factor is controversial. This study was performed to investigate the significance of lymphatic invasion as a prognostic factor in early results of gastric cancer after gastrectomy. We had been prospectively analyzed 362 consecutive patients resected from June 1994 to June 1996 in the Department of Surgery, Ajou University Hospital. Grading of the lymphatic invasion was determined by pathologists according to the JRSGC classification. The difference in the survival rate between the ly0 group and the ly1 group was statistically insignificant, so we only used ly2 and ly3 as the risk group. The prognostic values of lymphatic invasion as a single parameter for survival and early recurrence were determined by using univariate and multivariate analyses. Lymphatic invasion was observed in 90 (24.9%) of the 362 gastric cancer cases, and ly2 or ly3 were 66 (18.3%)cases. The incidence of ly> or =2 steadily increased with advancing TNM stage : stageI, 9/124 (7.3%); stageII, 5/50 (10.0%); stageIII, 24/111 (21.6%); stageIV, 28/78 (35.9%). The incidence of ly> or =2 was significantly correlated with other prognostic factors such as lymph node metastasis, the depth of the primary tumor, and the stage (p<0.05). The cummulative 2-year-survival rate determined by the Kaplan-Meirer method was 56.4% in patients with ly> or =2 and 90.9% in patients with ly<2. The log-rank test showed significant differences between the ly<2 group and the ly> or =2 group (p<0.05). Multivariate analysis of the survival function showed that the depth of the primary tumor had the highest prognostic value followed by lymphatic invasion (ly> or =2) as an independent prognostic factor (p=0.0008, Exp(B)=3.55). The cummulative recurrence free rate determined by the Kaplan-Meirer method was 78.7% in the ly> or =2 group and 90.54% in the ly<2 group. There was statistically significant difference (p<0.05). Univariate analysis for recurrence after radical resection showed that lymphatic invasion had a prognostic impact in the ly> or =2 group (p=0.0055). Therefore lymphatic invasion with ly2 or ly3 grade is an independent prognostic factor that is associated with poor prognosis for early results after gastrectomy. In grading of lymphatic invasion by gastric cancer, simple differentiation of whether lymphatic invasion is present or not was insignificant in the results of our study. In the future, the grade of lymphatic invasion in gastric cancer should be carefully classified by pathologists.