The aim of this study was to investigate the influence of soccer-specific fatigue on the kinematics of an agility sprint. Ten male professional soccer players (age 24.7±4.4 years, body mass 77.1±8.3 kg) completed an intermittent treadmill protocol replicating the activity profile of match-play, comprising two 45-min halves separated by a 15-min passive half-time interval. Pre-exercise and at 15-min intervals each player completed an agility sprint that consisted of a 180° cutting manoeuvre. Knee joint kinematics in the frontal and sagittal planes were determined for both the support and turning leg using a nine-camera automated motion analysis system operating at 200 Hz. During the penultimate foot contact, knee kinematics were characterized by joint flexion and increased varus alignment. Knee flexion at touchdown decreased significantly (P<0.05) as a function of exercise duration from 57.4±15.5° before exercise to 37.0±5.9° at the end of the second half. The range of joint movement during the knee flexion phase increased significantly during the first half (T 45=66.6±18.2°) and remained elevated during the second half (T 75=66.4±18.1°; T 90=65.7±20.4°; T 105=70.2±19.4°) relative to pre-exercise values (51.8±18.8°). During the final foot contact, knee kinematics were also characterized by flexion and increased varus alignment. Knee flexion at touchdown decreased during each half, with the knee angle at the end of the first half (30.6±7.0°) significantly (P=0.02) straighter than before exercise (39.5±6.3°), and significantly straighter at the end of the second half (30.2±2.9°) than after the half-time interval (37.7±7.8°) or before exercise. The range of knee flexion during ground contact increased significantly during each half. The range of knee varus during flexion changed from a varus displacement during the first 15 min to a valgus displacement thereafter. Peak valgus observed at the end of each half (T 45=4.7±7.9°; T 105=6.9±7.4°) was significantly (P<0.05) greater than before exercise. The range of valgus movement during knee extension was greatest following the passive half-time interval (T 60=6.2±7.3°), and tended to increase throughout the second half. Prolonged exposure to soccer-specific intermittent exercise therefore induced changes in knee kinematics that may have implications for injury incidence. The increased varus alignment and time-dependent decrease in knee flexion at touchdown represent two potential mechanisms for increased injury risk.