Familial adenomatous polyposis (FAP) provides a model for sporadic colorectal cancer development. Cyclooxygenase (COX) inhibition may ameliorate polyp development, but rofecoxib was withdrawn due to cardiovascular side effects. Although this selective COX-2 inhibitor, like diet, may alter the fatty acid and eicosanoid pattern, data on the potential alteration in tissues after use, are scarce. The aims were to study if rofecoxib might influence the fatty acid distribution in serum phospholipids and duodenal lesions, mRNA for COX-1 and COX-2 in leucocytes and duodenal lesions, and finally plasma levels of PGE(2) in a randomized, double-blind, placebo controlled study (n = 38). Significant reductions were found for essential fatty acid index both in serum phospholipids (P = 0.01, 95% CI = -0.9; -0.1), and in duodenal lesions (P = 0.04, 95 CI % = -0.9; -0.1) after treatment. No treatment effects were found on the COX mRNA expression, or in the plasma PGE(2) levels. Dietary AA/EPA ratio was inversely associated with all the indicators of EFA status (all P < 0.01). These findings suggest that the effects of COX chemoprevention should be further investigated in FAP and that dietary needs should be included in the treatment of FAP.