The Japanese randomized trial comparing standard D2 with D2 plus additional para-aortic lymph node (PAN) dissection for advanced gastric cancer (JCOG study 9501) did not demonstrate any difference in survival between the two groups. It is unknown whether there is any prognostic benefit in dissection for subgroups of PAN. Non-inferiority in survival of the patients with PAN metastasis to the patients having n2 metastasis was examined according to the subgroup of PANs and the tumor location. The survival curve of n2 patients (n=131) were retrospectively compared with that of patients with PAN metastasis (n=55) and also compared with that of patients with metastasis to subgroup of PANs by the location of primary tumor (regions U, M and L). Expectedly, the prognosis of the n2 patients is significantly better than that of the patients with PAN metastasis, but there was no difference in the survival times between the n2 (+) group and the a2-lat (+) or the b1-int (+) group, suggesting that the a2-lat or the b1-int dissection matched the D2 dissection. Furthermore, the importance in dissection of the a2-lat and the b1-int was investigated according to the primary tumor location. The patients with metastasis to a2-lat in the region U, a2-lat and b1-int in the region M and b1-int in the region L demonstrated prognostic non-inferiority to the patients having n2 metastasis. Selective lymphadenectomy of subgroups of PANs in which metastases are highly suspected according to the tumor location is one of treatment strategies to advanced gastric cancer.