Neuromuscular fatigue (NMF) is usually assessed non-invasively in healthy, athletic or clinical populations with the combination of voluntary and evoked contractions. Although it might appear relatively straightforward to magnetically or electrically stimulate at different levels (cortical/spinal/muscle) and to measure mechanical and electromyographic responses to quantify neuromuscular adjustments due to sustained/repeated muscle contractions, there are drawbacks that researchers and clinicians need to bear in mind. The aim of this opinion paper is to highlight the pitfalls inevitably faced when NMF is quantified. The first problem might arise from the definition of fatigue itself and the parameter(s) used to measure it; for instance, measuring power vs. isometric torque may lead to different conclusions. Another potential limitation is the delay between exercise termination and the evaluation of neuromuscular function; the possible underestimation of exercise-induced neural and contractile impairment and misinterpretation of fatigue etiology will be discussed, as well as solutions recently proposed to overcome this problem. Quantification of NMF can also be biased (or not feasible) because of the techniques themselves (e.g. results may depend on stimulation intensity for transcranial magnetic stimulation) or the way data are analyzed (e.g. M wave peak-to-peak vs first phase amplitude). When available, alternatives recently suggested in the literature to overcome these pitfalls are considered and recommendations about the best practices to assess NMF (e.g. paying attention to the delay between exercise and testing, adapting the method to the characteristics of the population to be tested and considering the limitations associated with the techniques) are proposed.