Non-steroidal anti-inflammatory drugs (NSAIDs) specifically inhibit cyclooxygenase (COX) activity and are widely used as antiarthritics, post-surgical analgesics, and for the relief of acute musculoskeletal pain. Recent studies suggest that non-specific KSAlDs, which inhibit both COX-I and COX-2 isoforms, delay bone healing. The objectives of this study were 2-fold; first, to measure the relative changes in the normal expression of COX-1 and COX-2 mRNAs over a 42 day period of fracture healing and second, to compare the effects of a commonly used non-specific NSAID, ketorolac, with a COX-2 specific NSAID, Parecoxib (a pro-drug of valdecoxib), on this process. Simple, closed, transverse fractures were generated in femora of male Sprague-Dawley rats weighing approximately 450 g each. Total RNA was prepared from the calluses obtained prior to fracture and at I , 3, 5, 7, 10, 14, 21, 35 and 42 days post-fracture and levels of COX-I and COX-2 mRNA were measured using real time PCR. While the relative levels of COX-1 inRNA remained constant over a 21-day period, COX-2 inRNA levels showed peak expression during the first 14 days of healing and returned to basal levels by day 21. Mechanical properties of the calluses were then assessed at 21 and 35 days postfracture in untreated animals and animals treated with either ketorolac or high or low dose parecoxib. At both 21 and 35 days after fracture, calluses in the group treated with the ketorolac showed a significant reduction in mechanical strength and stiffness when compared with controls (17 < 0.05). At the 21-day time point, calluses of the parecoxib treated animals showed a lower mean mechanical strength than controls, but the inhibition was not statistically significant. Based on physical analysis of the bones, 3 of 12 (25'%1) of the ketorolac-treated and 1 of 12 (8%) of the high dose parecoxib-treated animals showed failure to unite their fractures by 21 days, while all fractures in both groups showed union by 35 days. Histological analysis at 21 days showed that the calluses in the ketorolac-treated group contained substantial amounts of residual cartilage while neither the control nor the parecoxib-treated animals showed comparable amounts of cartilage at this stage. These results demonstrate that ketorolac and parecoxib delay fracture healing in this model, but in this study daily administration of ketorolac, a non-selective COX inhibitor had a greater affect on this process. They further demonstrate that a COX-2 selective NSAID, such as parecoxib (valdecoxib), has only a small effect on delaying fracture healing even at doses that are known to fully inhibit prostaglandin production.