To examine neuromuscular fatigue and recovery following an isokinetic fatiguing exercise, nine active females performed a fatiguing exercise comprising of ten sets of ten maximal concentric knee extensions. Before (pre-test), five minutes (post-test), 24 h and 48 h after the fatiguing exercise, maximal voluntary eccentric (-1.05 rad x s(-1); -2.09 rad x s(-1)), isometric (0 rad x s(-1)) and concentric (1.05 rad x s(-1); 2.09 rad x s(-1)) torque were measured. In order to distinguish central from peripheral factors involved in torque decrement, activation level (twitch interpolation technique) and twitch contractile properties were recorded. During the course of the fatiguing exercise, concentric torque was significantly lower during the 3rd set than pre-test (-5.6 +/- 12.3 %) and further decreased to the 10th (-10.3 +/- 9.5 %). Eccentric and isometric torques were significantly lower during post-test than pre-test (-16.8 +/- 8.8 % at -2.09 rad x s(-1), -15.1 +/- 7.4 % at -1.05 rad x s(-1), and -10.4 +/- 5.9 % at 0 rad x s(-1); p < 0.05), while concentric torque was not significantly modified. Voluntary activation, peak twitch torque, twitch maximal rates of force development and relaxation were also significantly declined (p < 0.05) at post-test. Twenty-four hours later, all the measured parameters were close to their pre-fatigue values. The present results reveal that the best way to test concentric-induced alteration of neuromuscular function was to use stressful testing conditions, such as eccentric contractions.