Cyclooxygenase-2 (COX-2) has been identified as a potential target for prevention and therapy of human colorectal cancers (CRC) and other cancers. Because right-sided colon cancers (RSCCs) exhibit clinicopathologic and genetic differences from left-sided colorectal cancers (LSCRCs), determination of COX-2 status in these subsets of CRCs may be clinically relevant in designing COX-2 inhibitor trials for CRC and in subsequent assessment of objective therapeutic response to such therapy. Thirty-six primary CRC resection specimens (18 left, 18 right) from 36 patients were evaluated. Representative tumor sections were subjected to immunohistochemical analysis of COX-2. A semiquantitative system was used to score cytoplasmic COX-2 immunostaining. The tumors were considered COX-2 positive if more than 10% tumor cells showed COX-2 staining. Clinicopathologic and COX-2 data were compared for LSCRCs versus RSCCs. All 18 LSCRCs and 13 of 18 (72%) RSCCs were well to moderately differentiated. Overall rates of COX-2 positivity for the LSCRCs versus RSCCs were 67% (12 of 18) and 33% (6 of 18), respectively (P = 0.04). Furthermore, 11 of 12 (92%) COX-2 positive LSCRCs and 3 of 6 (50%) COX-2 positive RSCCs were stage II-IV at resection. All 12 COX-2 positive LSCRCs were associated with advanced primary tumor and 4 of 12 (33%) LCRCs had distant metastases. These associations could not be evaluated for the RSCCs because of the limited number of COX-2 positive cases. The more frequent expression of COX-2 in LSCRCs as compared with RSCCs supports the hypothesis that COX-2 expression may be related to genetic alterations specific to right- or left-sided CRCs. Further studies are needed to elucidate such relationships. Our data also suggest that stratification of patients with CRC into right- and left-sided subsets may be important in optimal patient selection for COX-2 inhibitor therapy and for subsequent assessment of objective therapeutic response.