We have investigated aromatase and the inducible cyclooxygenase COX-2 expression using immunocytochemistry in tumors of a series of patients with advanced breast cancer treated with aromatase inhibitors. Aromatase was expressed in 58/102 breast cancers. This is similar to the percentage previously reported for aromatase activity. Interestingly, aromatase was expressed in a variety of cell types, including tumor, stromal, adipose, and endothelial cells. Since prostaglandin E2 is known to regulate aromatase gene expression and is the product of COX-2, an enzyme frequently overexpressed in tumors, immunocytochemistry was performed on the tissue sections using a polyclonal antibody to COX-2. Aromatase was strongly correlated (P<0.001) with COX-2 expression. These results suggest that PGE2 produced by COX-2 in the tumor may be important in stimulating estrogen synthesis in the tumor and surrounding tissue. No correlation was observed between aromatase or COX-2 expression and the response of the patients to aromatase inhibitor treatment. However, only 13 patients responded. Nine of these patients were aromatase positive. Although similar to responses in other studies, this low response rate to second line treatment suggests that tumors of most patients were no longer sensitive to the effects of estrogen. Recent clinical studies suggest that greater responses occur when aromatase inhibitors are used as first line treatment. In the intratumoral aromatase mouse model, expression of aromatase in tumors is highly correlated with increased tumor growth. First line treatment with letrozole was effective in all animals treated and was more effective than tamoxifen in suppressing tumor growth. Letrozole was also effective in tumors failing to respond to tamoxifen, consistent with clinical findings. In addition, the duration of response was significantly longer with the aromatase inhibitor than with tamoxifen, suggesting that aromatase inhibitors may offer better control of tumor growth than this antiestrogen.