Nonsteroidal anti-inflammatory drugs are commonly prescribed after rotator cuff repair. These agents can impair bone formation, but no studies have evaluated their impact on tendon-to-bone healing. Traditional nonselective nonsteroidal anti-inflammatory drugs and cyclooxygenase-2-specific nonsteroidal anti-inflammatory drugs interfere with tendon-to-bone healing. Controlled laboratory study. One hundred eighty Sprague-Dawley rats underwent acute rotator cuff repairs. Postoperatively, 60 rats received 14 days of celecoxib, a cyclooxygenase-2-specific nonsteroidal anti-inflammatory drug; 60 received indomethacin, a traditional nonselective nonsteroidal anti-inflammatory drug; and 60 received standard rat chow. Animals were sacrificed at 2, 4, and 8 weeks and evaluated by gross inspection, biomechanical testing, histologic analysis, and polarized light microscopy to quantify collagen formation and maturation. Five tendons completely failed to heal (4 celecoxib, 1 indomethacin). There were significantly lower failure loads in the celecoxib and indomethacin groups compared with the control groups at 2, 4, and 8 weeks (P < .001), with no significant difference between nonsteroidal anti-inflammatory drug groups. There were significant differences in collagen organization and maturation between the controls and both nonsteroidal anti-inflammatory drug groups at 4 and 8 weeks (P < .001). Controls demonstrated progressively increasing collagen organization during the course of the study (P < .001), whereas the nonsteroidal anti-inflammatory drug groups did not. Traditional and cyclooxygenase-2-specific nonsteroidal anti-inflammatory drugs significantly inhibited tendon-to-bone healing. This inhibition appears linked to cyclooxygenase-2. If the results of this study are verified in a larger animal model, the common practice of administering non-steroidal anti-inflammatory drugs after rotator cuff repair should be reconsidered.