Abstract: Many experimental researches on the in-plane random stacked graphene composites (GC) for wearable sensors have been carried out. Due to the limitation of experimental technology, its piezoresistive sensing mechanism and piezoresistive performance are still an open problem. Based on the microstructure characteristics of GC, the position and direction of graphene flakes in GC are determined by the uniformly distributed random numbers between 0 and 1, and a novel two-dimensional GC model is established. According to the uniform deformation characteristics of GC, the finite element method for piezoresistive performance of GC is developed. The relative resistance, gauge factor, morphology of graphene flakes and current density contour of GC are obtained. By connecting them, it is revealed that the substantial cause of piezoresistive effect is the change of graphene flakes density and the specific reason is the variation of the electron migration pathway and the invalid flakes number. The increase of the length of electron migration pathway caused by the relative sliding of overlapped graphene flakes leads to the linear sensing characteristics, while the cut of the number of electron migration pathway and the increase in the number of invalid flakes induced by the separation of graphene flakes bring about the non-linear sensing effect. In addition, the results show that GC with high area fraction and GC with large-scale graphene flakes have a large sensing range, while GC with low area fraction and GC with small-scale graphene flakes have a higher gauge factor. Finally, the in-plane resistivity of contact surface of overlapped graphene flakes is assumed as a function of strain and the influence of contact resistance on piezoresistive effect of the GC is investigated, to understand the mechanism of the GC piezoresistive performance. These results can contribute to the improvement or innovation of GC fabrication method and the production of the GC piezoresistive sensing devices for expected sensing performance.