Several studies have indicated that administration of non-steroidal anti-inflammatory drugs (NSAID) to patients with familial adenomatous polyposis (FAP) results in a regression of colorectal adenomas through inhibition of cyclooxygenase-2 (COX-2). It is thought that sporadic colorectal adenomas might also be useful targets for the chemoprevention of colorectal cancer, but a marked effect of NSAID on the regression of sporadic adenomas has not been observed. We investigated the immunohistochemical expression of COX-2 in sporadic tubular adenomas (n = 100) from 63 patients and in tubular adenomas (n = 121) from 12 patients with FAP, in order to determine if chemoprevention might be more successful in sporadic adenomas once they have reached a certain size. COX-2 scores were significantly lower (P < 0.0001) in small (< 5 mm in diameter) adenomas than in large (> or = 5 mm) adenomas. This was observed in both sporadic cases and in cases involving patients with FAP. With regard to small (< 5 mm) adenomas, significantly higher (P = 0.02) COX-2 scores were obtained in adenomas resulting from FAP than sporadic adenomas. The variation in COX-2 expression observed among sporadic adenomas of different sizes should be taken into account when making decisions regarding attempts at chemoprevention using NSAID. Sporadic adenomas 5 mm or larger with upregulated COX-2 expression are potentially useful targets for the antiproliferative effects of NSAID.