Aim: To explore the feasibilty of performing minimally invasive surgery (MIS) on subsets of submucosal gastric cancers that are unlikely to have regional lymph node metastasis. Methods: A total of 105 patients underwent radical gastrectomy with lymph node dissection for submucosal gastric cancer at our hospital from January 1995 to December 1995. Besides investigating many clinicopathological features such as tumor size, gross appearance, and differentiation, we measured the depth of invasion into submucosa minutely and analyzed the clinicopathologic features of these patients regarding lymph node metastasis. Results: The rate of lymph node metastasis in cases where the depth of invasion was <500 μm, 500-2 000 μm, or >2 000μm was 9% (2/23), 19% (7/36), and 33% (15/46), respectively (P<0.05). In univariate analysis, no significant correlation was found between lymph node metastasis and clinicopathological characteristics such as age, sex, tumor location, gross appearance, tumor differentiation, Lauren's classification, and lymphatic invasion. In multivariate analysis, tumor size (>4 cm vs ≤2 cm, odds ratio = 4.80, P = 0.04) and depth of invasion (>2 000 μm vs ≤500 μm, odds ratio = 6.81, P = 0.02) were significantly correlated with lymph node metastasis. Combining the depth and size in cases where the depth of invasion was less than 500 μm, we found that lymph node metastasis occurred where the tumor size was greater than 4 cm. In cases where the tumor size was less than 2 cm, lymph node metastasis was found only where the depth of tumor invasion was more than 2 000 μm. Conclusion: MIS can be applied to submucosal gastric cancer that is less than 2 cm in size and 500 μm in depth.