The objective of the present study was to investigate the possible relationship between bispectral parameters extracted from surface EMG (sEMG) signals and muscle force and fatigue. Our hypothesis was that changes in motor unit recruitment during muscle contraction and fatigue, affect sEMG distribution and the degree of complexity and irregularity in the muscle. Thus, four features based on higher order spectra and cumulants were extracted from sEMG signal, recorded from biceps brachii muscle of a healthy female volunteer during rest, sustained (fatiguing) 50% MVC, 100% MVC and recovery. Results obtained from weighted center of bispectrum (WCOB) analysis showed that the values of f1m and f2m were higher during rest and recovery states, while they decreased during MVCs. However, when fatigue occurred, these parameters increased slightly, again. Moreover, entropy features, namely NBE and NBSE decreased with contraction compared to rest and recovery states, indicating less complexity of time series during MVCs. However, the changes were not significant during fatigue and during changes in MVC levels from 50% to 100%. On the other hand, test of non-Gaussianity based on negentropy showed the reverse pattern of WCOB, NBE and NBSE. In addition, contour maps of bispectrum enabled us to visually differentiate each trial.