The purpose of this study was to determine the reliability of an isokinetic endurance task. After habituation, 10 physically active males completed three trials of an isokinetic endurance task on nonconsecutive days. The task was performed on a Cybex Norm dynamometer. It consisted of 30 maximal, concentric and eccentric, contractions of the hip extensors at 180°/s. Reliability was assessed through technical error of measurement (TEM), relative TEM, group coefficient of variation and intraclass correlation (ICC). During eccentric actions both average power per repetition (APR) and work done (WD) showed very good reliability. For APR the group coefficient of variation was 8.0% and the ICC was 0.93. The TEM and relative were 28.9 W and 16.3%, respectively. For WD these values were 5.6%, 0.94%, 451% and 12.4% J, respectively. By contrast, fatigue index (FI) demonstrated poor reliability with TEM, group coefficient of variation and ICC values of 16.2%, 26.1% and 0.32%, respectively. The reliability from the concentric actions demonstrated more variable results. The group coefficient of variation was low for both APR (5.9%) and WD (5.3%), however, the ICC values were lower than for eccentric actions with a score of 0.85 for both APR and WD. Once again FI demonstrated low reliability with a TEM, group coefficient of variation of 12.8%, 28.3% and ICC of 0.72%. The reliability of the hip extensors, both concentrically and eccentrically, depended on the dependent variable measured. FI showed poor reliability, while WD showed high reliability.