Mechanical loading is known to increase connective tissue blood flow of human tendons and to cause local release of vasodilatory substances. The present study investigated the importance of prostaglandins (PG) formed by cyclo-oxygenase isoforms (COX-1 and 2) for the exercise-related increase in blood flow in connective tissue. Healthy individuals (n = 24, age: 23-31 years) underwent 30 min of intermittent, isometric, plantarflexion with both calf muscles either without (n = 6, Control, C) or with blockade of PG formation, either COX-2 specific (n = 10, Celecoxib 2 x 100 mg day-1 for 3 days prior to the experiment) or COX unspecific (n = 8, indomethacin 100 mg (12 and 1 h pre-experiment) and acetyl salicylic acid 500 mg day-1 for 3 days pre-experiment). Prostaglandin E2 (PGE2) concentration was determined by microdialysis and blood flow by 133Xe washout. In C, interstitial PGE2 rose from (0.8 +/- 0.2 (rest) to 1.4 +/- 0.5 ng ml-1 (exercise), P <0.05), whereas during unspecific COX inhibition, tissue PGE2 was completely inhibited at rest and during exercise. COX-2 specific blockade did not inhibit tissue PGE2 at rest, but totally abolished the exercise induced increase. Blood flow was similar in the three groups at rest (P > 0.05), whereas the increase in flow with exercise was reduced by 35 and 43 % with COX-2 specific blockade (3.2 +/- 0.7 to 6.1 +/- 1.5 ml (100 g tissue)-1 min-1 or COX unspecific blockade (3.0 +/- 0.8 to 7.6 +/- 1.6), respectively, compared to C (2.7 +/- 0.8 to 10.2 +/- 2.0)(P <0.05). The findings indicate that COX-2 specific mechanisms are responsible for the exercise-induced increase in prostaglandin synthesis, and that increase in tissue prostaglandin plays an important role for blood flow in peritendinous connective tissue during physical loading in vivo.