Four hundred and eighty five gastric cancer patients underwent gastrectomy in the past 6 years and of those 332 patients (68.2%) were found to have only serosal involvement (se) or adjacent organ invasion (sei). Simple gastrectomy (SG) was carried out in 144 cases while radical gastrectomy (RG), which included systemic lymphadenectomy, was performed in 188 cases. This type of operation was chosen by the operators arbitrarily except in the presence of non-curative factors such as multiple liver metastasis, peritoneal seeding and distant metastasis, in which case SG was always used to relieve gastrointestinal obstruction or bleeding. The operative mortality of SG was 2.7% and that of RG was 3.2%. On average, 15.8 lymph nodes were taken out per patient in SG and 33.6 lymph nodes in RG. The chance of nodal metastasis was higher in sei than in se. According to the Japanese General Rules of Gastric Cancer Study, we found that for gastric cancer with serosal involvement: (1) the long term result of RG was superior to SG in both stage 3 and 4 patients, (2) when absolute curative resection was achieved, the result was the best in stage 3 patients, (3) no patient in stage 4 who underwent absolute non-curative resection survived for more than 42 months, therefore this result was the worst. RG is recommended for advanced gastric cancer patients with serosal involvement whenever the patient's condition permits and in absence of the unresectable non-curative factors.