PURPOSE: Because depth of invasion by T3 rectal cancer can vary according to the extent of mesorectal invasion, the prognosis for invasive T3 rectal cancer is reported to be very different from that for minimal invasive cancer. Recently, with more emphasis on circumferential resection margin (CRM) status, the T stage, rather than the N stage, seems to be a more valuable prognostic marker in rectal cancer. Therefore, the aim of this study is to determine the prognostic significance of the CRM in invasive T3 rectal cancer. METHODS: Through reviewing 324 consecutive patients with rectal cancer who underwent a curative resection between January 1995 and December 2002 at Busan Paik hospital, 195 patients with invasive T3 rectal cancer, who had not received preoperative neoadjuvant therapy were selected. The patients were classified into a negative CRM group (negative group, n=173) or a positive CRM group (positive group, n=22), and the patients were subgrouped according to the presence of lymph-node (LN) metastasis and CRM status as negative LN and negative CRM (L-/CM-), negative LN and positive CRM (L-/CM+), positive LN and negative CRM (L+/CM-) and positive LN and positive CRM (L+/CM+). All pathological specimens were re-reviewed by a single pathologist, and the distance between the most advanced edge and the outermost aspect of the specimen was re-measured by using a microscope. Local relapse rates, disease free survival, and overall survival were compared using the Kaplan- Meier method. Multivariate analyses to identify independent prognostic factors were performed using the logistic regression model. RESULTS: Local recurrence rates in the positive group and the negative group were 38.6% and 15.3%, respectively (P=0.004, log-rank test). The multiple logistic regression model demonstrated positive CRM (hazard ratio 4.4, P=0.0007) and N2 nodal status (hazard ratio 2.4, P=0.02) as predictors of local recurrence. In the subgroup analysis, the overall recurrence rates and survival rates were, respectively, 12.3% and 86.5% in the L-/CM- subgroup, 53.1% and 50.3% in the L-/CM+ subgroup, 52.7% and 50.0% in the L+/ CM- subgroup, and 58.7 % and 33.8% in the L+/CM+ subgroup (log rank test for trend; P=0.0001 and P=0.0001, respectively). CONCLUSIONS: In the event of predicted CRM involvement in invasive T3 rectal cancer, adjuvant therapy should be performed to improve local control. Also, larger prospective studies are needed to clarify the prognostic role of the CRM in invasive T3 rectal cancer because the number of cases in this study was small, especially in the number of CRM positive cases.