There is no direct evidence to support the contention that contracting skeletal muscle and/or associated vasculature generates free radicals in exercising humans. The unique combination of isolated quadriceps exercise and the measurement of femoral arterial and venous free radical concentrations with the use of electron paramagnetic resonance (EPR) spectroscopy enabled this assumption to be tested in seven healthy men. Application of ex vivo spin trapping using alpha-phenyl-tert-butylnitrone (PBN) resulted in the detection of oxygen- or carbon-centered free radicals (a(N) = 1.38 +/- 0.01 mT and a(beta)(H) = 0.17 +/-0.01 mT, where a(N) and a(beta)(H) are the nitrogen and beta-hydrogen coupling constants, respectively) with consistently higher EPR signal intensities of the PBN spin adduct observed in the venous compared with the arterial circulation (P < 0.05). Incremental exercise further increased the venoarterial intensity difference [85 +/- 58 arbitrary units (AU) at 24 +/- 6% maximal work rate (WR(max)) vs. 387 +/- 214 AU at 69 +/- 7% WR(max); P < 0.05]. When combined with measured changes in femoral venous blood flow (Q), this resulted in a net adduct outflow of 130 +/- 118 and 1,146 +/- 582 AU/min (P < 0.05), which was positively associated with leg oxygen uptake (r(2) = 0.47, P < 0.05) and Q (r(2) = 0.47, P < 0.05). These results provide the first evidence for oxygen- or carbon-centered free radical outflow from an active muscle bed in humans.