Traditional electromyographic (EMG) measures (e.g., amplitude, mean and median frequencies of the power spectra) have demonstrated inconsistent abilities in monitoring localized muscle fatigue at relatively low effort levels. In the present study, several alternative EMG-based fatigue indices were evaluated, derived using a logarithmic representation of the power spectrum, the fractal dimension of the raw signal, and a Poisson distribution fit to the power spectrum. These methods, along with traditional approaches, were applied to EMG data obtained from three separate experiments. In the first two experiments, 24 participants performed sustained isometric shoulder abductions and torso extensions at 30% of maximum voluntary strength (MVC). In the third experiment, another group of 12 participants performed similar shoulder exercises at 15% and 30% MVC, with repeatability assessed at 15% MVC. Both traditional and alternative EMG measures were analyzed for their 'utility', in terms of sensitivity to fatigue, variability, repeatability, and predictive ability. Results demonstrated that parameters derived from fractal analysis and the Poisson distribution demonstrated high utility. These alternative approaches appear promising as fatigue indices for low level isometric tasks.