OBJECTIVE: To evaluate the independent prognostic ability of the American Joint Committee on Cancer (AJCC) tumor regression scores within pathologic stage II and III rectal cancers. BACKGROUND: Response to neoadjuvant chemoradiation (nCRT) has been debated as a biologic surrogate for tumor biology and prognosis in rectal cancer. AJCC regression scores have been shown to correlate with prognosis. METHODS: Patient demographics, tumor characteristics, and AJCC scores (0 = complete response; 1 = isolated tumor cells remaining; 2 = residual cancer outgrown by fibrosis; 3 = extensive residual cancer) were assessed from 545 rectal cancer patients treated by nCRT followed by surgery at a single institution. Patients were classified as responders (score 0-2) or nonresponders (score 3). Survival analyses were performed using Cox proportional hazards models. RESULTS: Of 545 cases, 123 and 182 were pathologic stage II and III, respectively. Median follow-up was 4.9 years. AJCC regression scores were not independently prognostic within stage II cancers. However, AJCC scores were strongly associated with prognosis within stage III cancers (nonresponse 5-year overall survival [OS] 27% vs 67%, P < 0.001). Stage III responders (N = 139, 76.4%) had similar outcomes to stage II (5-year OS 67% vs 74%, P = 0.89). Conversely, stage III nonresponders (N = 43, 23.6%) approached stage IV outcomes (5-year OS 27% vs 18%, P = 0.09). On multivariable analysis, nonresponse (hazard ratio 3.2, 95% confidence interval 1.7-6.2), along with positive margin, abdominoperineal resection, and no adjuvant chemotherapy administration were independently associated with worse OS. CONCLUSIONS: AJCC response score after nCRT is a novel prognostic factor in pathologic stage III rectal cancer and may guide surveillance and adjuvant therapy decisions.