BACKGROUND/AIMS: Since surgical results in advanced gastric cancer remain poor and para-aortic lymph node dissection may contribute to survival, it is useful to determine the significance of para-aortic lymph node dissection. METHODOLOGY: Para-aortic lymph node dissection was provisionally indicated for patients with invasion depth deeper than the subserosal layer. Clinicopathologic variables were retrospectively analyzed using univariate analysis and multivariate analysis to predict para-aortic lymph node metastasis. Similarly, they were analyzed using univariate analysis and the Cox's proportional hazards regression model to estimate the prognostic factor in 120 patients who underwent para-aortic lymph node dissection. Surgical results and post-operative complications were compared between para-aortic lymph node dissection and D2 dissection. RESULTS: Univariate analysis revealed that the mean diameter, the degree of lymph node metastasis, and the invasion depth were significant predictors of para-aortic lymph node metastasis. Multivariate analysis showed that n2 was the only independent predictive factor as to para-aortic lymph node metastasis. Univariate analysis revealed tumor site, tumor diameter, lymph node metastasis, number of positive lymph nodes, INF, and stage were significantly associated with 5-year survival. The Cox's proportional hazards regression model showed that the number of positive lymph nodes and the number of positive para-aortic lymph nodes were independent prognostic factors. Patients with < or = 10 positive lymph nodes in any stage or < or = 3 positive para-aortic lymph nodes in stage IVb had significantly better surgical results. Surgical results for patients who underwent para-aortic lymph node dissection with n2 or invasion depth deeper than the exposed serosa were significantly higher than those in D2. As to post-operative complications, pancreatic fistula and respiratory complications were significantly frequent after para-aortic lymph node dissection. CONCLUSIONS: n2 is helpful in predicting para-aortic lymph node metastasis. Whereas, post-operative morbidity such as pancreatic fistula and respiratory complications after para-aortic lymph node dissection were significantly higher, they were controllable. Para-aortic lymph node dissection should be indicated in advanced gastric cancer patients in which lymph node metastasis is over n2 or invasion depth is deeper than the exposed serosa. But the number of positive para-aortic lymph nodes must be less than three.