Prolonged performance of repetitive movements can lead to muscle fatigue, negatively impacting human performance. As a result, researchers have explored methods to effectively assess and quantify this phenomenon, where surface electromyography (sEMG) is a popular method to reveal information regarding muscle contractions. The continuous wavelet transform (CWT) captures the instantaneous frequency components of signals, which make it suitable for sEMG analyses of dynamic muscle contractions. Moreover, sample entropy (SampEn) can be used to quantify the complexity of the sEMG signal, which provides novel insights for assessing muscle fatigue. However, the amount of research on sEMG complexity analyses to assess muscle fatigue during dynamic contractions is limited. Therefore, the goal of this work was to: 1) calculate and compare the major frequency components (MFC) from CWT and modified SampEn (MSE) of sEMG signals during a repetitive trunk flexion-extension (F-E) task; and 2) determine which sEMG metric is more closely related to ground truth fatigue indicators including the visual analogue scale (VAS), maximum pulling force, and kinematic variability of movements. Seven male and five female participants performed up to twelve sets of 50 repetitive trunk FE movements based on pre-defined stopping criteria. Their VAS and maximum pulling strength were measured immediately after each set. The MFC from CWT and the MSE values were calculated from both the left and the right lumbar erector spinae (LES) throughout the movements. Trunk dynamic kinematic variability of every set was quantified by the spine motion composite index (SMCI). Repeated measures correlation coefficients (r) were used to calculate the relationship between MFC and MSE, as well as between these outcome variables and VAS, maximum pulling force, and SMCI across all participants. Visual inspection revealed that on average that both the MFC and the MSE of sEMG signals decreased as the fatiguing protocol progressed, where a significant correlation was found between the two sEMG metrics (r = 0.270, p = 0.006). No significant correlations were found between the two sEMG measures and the maximum pulling strength (r_MFC = 0.101, p = 0.313; r_MSE = 0.193, p = 0.051). Nevertheless, both sEMG metrics showed significant correlations with fatigue VAS, with the MFC having stronger correlations across all the participants (r_MFC = −0.602, p < 0.001) than the MSE (r_MSE = −0.248, p = 0.011). Significant negative correlations were also observed between the SMCI and both sEMG MFC (r_MFC = −0.268, p = 0.010) and MSE (r_MSE = −0.335, p = 0.001). Both sEMG metrics mapped onto the perceived fatigue and movement pattern variations during the task, suggesting they could be used for assessing fatigue during dynamic movements. However, the MFC had a stronger correlation with participants' perceived fatigue whereas MSE was more strongly correlated with kinematic variability. Continued research is required to further examine these relationships, as well as determine the best method of assessing changes in force output with muscle fatigue.