OBJECTIVE Lymphatic spread is an important prognostic factor in gastric cancer. The TNM classification requires at least 15 lymph nodes to stage and identify three prognostic groups according to the number of metastatic lymph nodes: N1 (1-6), N2 (7-15), N3 (> 15). The aim of this study was to investigate which type of lymph node dissection allows an accurate staging. METHODS From 1996 to 2001, we treated 140 gastric cancer patients, 27 with D1 and 113 with D2 dissection. We evaluated lymph node count, status and ratio between metastatic and total number of excised lymph nodes, keeping 20% as the cutoff value. RESULTS The mean number of lymph nodes was 18 and 33 respectively for D1 and D2 (P < 0.001), 41% of patients in D1 and 5% in D2 had less than 15 lymph nodes (P < 0.001). 59% in D1 and 73% in D2 (P = 0.145) had lymph node metastases, but this incidence decreased to 36% (P = 0.045) and 16% (P < 0.001) respectively for D1 and D2 when less than 15 lymph nodes were available. Considering the ratio between metastatic and total number of lymph nodes, 45% of D1 versus 3% of D2 (P < 0.001) in the N1 group exceeded 20%. CONCLUSIONS D2 lymph node dissection is better than D1 in providing at least 15 lymph nodes required for a correct staging. We confirm the risk of a downstage when less than 15 lymph nodes are available.