Lymphatic invasion is known as an independent predictor of lymph node metastasis in gastric cancer. However, the diagnosis of lymphatic invasion is sometimes difficult by hematoxylin-eosin (H&E) staining. Immunostaining using D2-40 was performed to study the distribution of lymphatic vessel and lymphatic invasion in a series of 78 primary gastric cancers. D2-40 showed specific staining for the lymphatic vessels, but not for blood vessels. The lymphatic invasion was most frequently found in the upper half of submucosal layer. Positive rate of lymphatic invasion by H&E staining was 27% (21/78), and that by D2-40 was 44% (34/78). Lymphatic invasion on H&E staining was diagnosed as false negative in 17 (21.8%) of 78 primary gastric cancers and false positive in 4 (5.1%) of 78 primary gastric cancers. Sensitivity for lymph node metastasis by the lymphatic invasion diagnosed by D2-40 was significantly higher (89%, 24/27) than by H&E staining (41%, 11/27). These results suggest that the diagnosis of lymphatic invasion by D2-40 is more sensitive than H&E staining. Sensitivity for the prediction of lymph node metastasis from the lymphatic invasion status in primary tumor by D2-40 was significantly higher than by H&E staining. Based on our results, we recommend the use of D2-40 immunoreactions for the routine evaluation of lymphatic invasion in gastric cancer.