OBJECTIVE: To characterize the reproducibility of three indicators of fatigue (FIs) specific to concentric plantar-flexion isokinetic testing at knee flexion (KF) angles of 0 degrees (straight) and 45 degrees (bent). METHODS: On two separate days one week apart, thirty-four males performed 50 consecutive maximal concentric isokinetic plantar-flexion contractions at 60 degrees/s with 0 degrees and 45 degrees of KF. Differences in the pre- and post-maximal voluntary isometric contraction torques (FIstatic), powers during the initial five and last five isokinetic contractions (FIdynamic) and powers during the 50 isokinetic contractions (FIslope) were used as FIs. Changes in means, intra-class correlation coefficients (ICCs) and coefficients of variations (CVs) were computed to quantify the reproducibility of the FIs. Comparisons were made between the two KF angles and three FIs using two-way repeated measures ANOVA. RESULTS: For both KF angles and three FIs, ICCs ranged from 0.52 to 0.71 and CVs from 10.0 to 29.3%. The CVs from the two isokinetic-based FIs were lower than those from the isometric FI and a trend towards larger ICCs at 0 degrees was observed. CONCLUSIONS: The reproducibility of the three FIs was not sufficient to be termed 'acceptable'. The FIslope and FIdynamic were more reproducible than FIstatic and are recommended - with the knee straight rather than bent - until more reliable indicators become available.