Purpose: Despite knowledge of the adverse effects of resection-line disease, surgeons continue to perform inadequate resections. This demonstrates the need for a more aggressive approach to assessment of resection margins at operation. Materials and Methods: Seven hundred fifteen gastric cancer patients who were operated on at our hospital from 1992 to 1998 were included in this analysis. Various clinicopathological factors, including resection-line involvement, were ascertained from the surgical and histopathological records. Results: Of the 715 evaluable patients, 27 patients ($3.8\%$) had involvement of one or both resection lines; in 10 patients the proximal resection line only, in 16 the distal resection line only, and 1 both resection lines were involved. Presence of resection-line involvement was significantly associated with T3 and T4 stage, N (+) stage, M (+) stage, type of operation (total gastrectomy), tumor location (entire stomach), size$geq$11 cm), and gross type of tumor (Borrmann 4 type). When performing a distal subtotal gastrectomy, no involvement was found when the cranial and caudal distances between the lesion and the line of transection was equal to or greater than 2 cm and 3 cm, respectively, for early cancer and 7 cm and 3 cm, respectively, for advanced cancer. When performing a total gastrectomy for upper 1/3 or middle 1/3 gastric cancer, no involvement was found when the cranial distances between the lesion and the line of transection were equal to or greater than 3 cm and 4 cm, respectively, without distinction of the presence of serosal invasion. Conclusions: The difference in survival between positive and negative margin patients is limited to the group of patients with curative surgery. An important principle of treatment is that the entire tumor must be removed with a 3 cm distal margin and a 2- to 7 cm margin depending on the location and the depth of wall invasion of the tumor, to provide histologically negative margins.