This study investigated the hypothesis that fatiguing exercises led to increased force fluctuations during submaximal isometric knee extensions and to decreased accuracy and steadiness in the time and frequency domains. Sixteen young adults (eight males, eight females) were tested, in a seated posture with 90°knee flexion, to assess their ability to reproduce target extensor torques set at 15 per cent and 20 per cent of their maximum voluntary isometric contraction, both before and after fatiguing exercises. Normalized mean (NMAE) and peak (NPAE) of the absolute error were both used to quantify accuracy, whereas normalized standard deviation of the absolute error (NSAE) was used to quantify steadiness of the torque trials in the time domain. Mean and median power frequencies (MnPF, MdPF) and normalized peak power (NPkP) were used to assess the spectral structure of the torque signals. NMAE, NSAE and NPAE all showed excellent intra-as well as intersession reliabilities (intraclass correlation values greater than 0.75 and low standard error of measurement values), demonstrating repeatability of the test set-up. NMAE, NSAE and NPAE increased significantly postfatigue (greater than 42%), together with a shift towards higher frequency (MnPF and MdPF) components, indicating that the set-up was sensitive enough to detect the decreased force accuracy and steadiness of the musculature after fatigue. Increased force variability in both the time and frequency domains could therefore explain decreased steadiness after fatigue.