PURPOSE Laboratory and case-control studies suggest a pivotal role for the cyclooxygenase-2 (COX-2) pathway in colorectal carcinogenesis. The purpose of this study was to test whether the COX-2 inhibitor rofecoxib could reduce recurrence and improve survival when administered in the adjuvant setting of colorectal cancer (CRC). PATIENTS AND METHODS Patients who had undergone potentially curative surgery and completion of adjuvant therapy for stage II and III CRC were randomly assigned to receive rofecoxib (20 mg daily) or placebo. The primary end point was overall survival (OS). Where formalin-fixed paraffin-embedded tumor tissue samples were available, COX-2 expression was evaluated by immunohistochemistry and correlated with clinical outcome. Results Two thousand four hundred thirty-four patients were entered onto the study. The trial was terminated early because of the worldwide withdrawal of rofecoxib. At this point, 1,167 patients had received rofecoxib and 1,160 patients had received placebo for median treatment durations of 7.4 and 8.2 months, respectively. For the rofecoxib and placebo arms, median follow-up times were 4.84 and 4.85 years, with 241 and 246 deaths and 297 and 329 recurrences, respectively. No difference was demonstrated in OS (hazard ratio [HR] = 0.97; 95% CI, 0.81 to 1.16; P = .75) or recurrence (HR = 0.89; 95% CI, 0.76 to 1.04; P = .15) comparing the two groups. Tumor COX-2 expression by immunohistochemistry was assessed for 871 patients, but neither prognostic nor predictive effects were observed. CONCLUSION In this study of abbreviated therapy in the adjuvant setting of CRC, rofecoxib did not improve OS or protect from recurrence in unselected patients. In addition, COX-2 expression did not correlate with prognosis overall or predict effectiveness of COX-2 inhibitors.