In vivo, indomethacin blockade of bone formation has been used to illustrate the role of prostaglandins. Indomethacin blocks the constitutive (COX‐1) and inducible (COX‐2) forms of cyclo‐oxygenase, and is therefore nonspecific in its action. To test the hypothesis that COX‐2 mediates the bone formation response to loading, rats were treated with vehicle, NS‐398 (a specific COX‐2 inhibitor) or indomethacin at 0.02, 0.2, or 2.0 mg/kg p.o. 3 h before loading the right tibia in four‐point bending. Bending or sham loads of 65 N were applied for one bout of 300 cycles and bone formation assessed 5–8 days after loading. Mechanically induced bone formation at the endocortical surface was calculated by subtracting formation indices of the left leg (control) from those of the right (loaded), and woven bone surface and area were measured at the periosteal surface. Endocortical bone formation was significantly increased by bending but not sham loading (p < 0.05). The increase in the endocortical bone formation rate and mineralizing surface caused by bending was only partially inhibited by indomethacin, even at the highest dose, whereas NS‐398 completely blocked bone formation at all doses (p < 0.05). The mineral apposition rate was depressed in a dose‐response fashion by NS‐398 (p < 0.05), but not by indomethacin. Woven bone formation at the periosteal surface was not prevented by treatment with indomethacin nor NS‐398, suggesting that its formation is not dependent on prostaglandin production. These data suggest that induction of COX‐2 is important for lamellar bone formation elicited by mechanical strain. (J Bone Miner Res 1996;11:1688‐1693)