The aim of this work is to perform a sensitivity analysis of a high density surface electromyogram (HD-sEMG) amplitude descriptors according to several grid parameters. This study is motivated by the fact that the electrode grid position and layout are crucial to record relevant electromyographic data. For this purpose, an analytical limb model is used, where the upper limb is modeled as a multilayered cylinder with three layers: muscle, fat tissue and skin tissue. Using this model, HD-sEMG signals are computed over the skin as a 2D surface along angular and longitudinal directions. Electrode recording is performed through a surface integration on the 2D surface according to the electrode shape. 16 simulations on 10 anatomies (350 Motor Units) with the same parameters were computed for 3 constant contraction levels: 30%, 50% and 70% of the Maximal Voluntary Contraction (MVC). Then, a global sensitivity analysis using the Elementary Effect Method is performed to explore the sensitivity of amplitude descriptors (Average Rectified Value, Root Mean Square value and High Order Statistics) relative to varying parameters from the electrode grid (inter-electrode distances, electrodes radius, position and rotation). From those grid definitions, monopolar, bipolar and laplacian signals are also computed to see the electrode arrangement sensitivity. The obtained results exposed a huge impact of the grid rotation on the studied criteria. They also showed that parameters specific to the electrode grid layout (inter-electrode distances) have the less impact. Moreover, they exhibited the laplacian arrangement as the most sensitive electrode arrangement to grid modifications.