After radical surgery for rectal adenocarcinoma, the presence of venous and neural invasion of tumor cells was correlated with the pattern of treatment failure, local in the pelvis or distant. Of 128 operation specimens, venous and neural invasion was demonstrated in 22 percent and 32 percent, respectively. A significant decrease of the distant recurrence-free 5-year survival (Kaplan-Meier method) was seen when venous invasion was demonstrated (32.9 percent vs. 84.3 percent; P less than 0.0001), whereas more local failures were registered in patients with neural invasion. The local recurrence-free 5-year survival in patients with neural invasion was 64.3 percent, compared with 81.1 percent when neural invasion was not demonstrated (P = 0.03). Their prognostic value was then studied in a Cox regression model including stage and grade. Neural invasion had the strongest association with local recurrences, whereas venous invasion was found to be the third strongest independent predictor of metastasis, after lymph node status and extent of local tumor infiltration. We conclude that examining for the presence of venous and neural invasion gives reliable prediction of recurrences after radical resection of rectal cancer. Recording of tumor recurrence pattern may lead to a better selection of patients for adjuvant therapy after surgery.