The distribution of lymph node metastasis and the clinicopathologic risk factors for nodal involvement in ovarian carcinoma need to be clarified based on systematic lymph node dissection. We studied 115 patients with ovarian carcinoma who underwent systematic pelvic and para-aortic lymph node dissection between 1987 and 1997. The incidence and distribution of lymph node metastasis are described and the clinico-pathologic risk factors for nodal involvement are investigated. Based on the occurrence of lymph node metastasis in the early stages, the incidence of solitary node involvement and the distribution of lymph node metastasis, we conclude that the primary site of nodal involvement in ovarian carcinoma is the para-aortic node (PAN), especially PAN superior to the inferior mesenteric artery (IMA). By univariate analysis, clinical stage, histologic type (mucinous vs. others), grade, multiple peritoneal metastases, peritoneal cytology, volume of ascites and serum CA125 level were correlated with overall incidence of lymph node metastasis. By performing a multivariate analysis with the clinical stage excluded, it was revealed that grade and peritoneal cytology were independent factors for PAN metastasis (p < 0.0025 and < 0.001, respectively) and that multiple peritoneal metastases and PAN metastasis were significant predictors of pelvic node metastasis (p < 0.01 and < 0.005, respectively). In conclusion, the PANs superior and inferior to IMA should be explored in staging of ovarian carcinoma that appears to be confined to the ovaries. To determine accurately the extent of disease, both the para-aortic and pelvic areas may need to be sampled or dissected in the case of ovarian carcinoma involving the peritoneal surfaces.