This work presents a model that allows to explain the resistance variability of OxRAM devices, both in high and low resistive states. This model is based on the calculation of a 3D resistance network, using trap assisted tunneling current. The stochastic nature of the resistance is captured by random placement of traps within a specific spatial distribution. The model is able to capture both cycle-to-cycle (temporal) and device-to-device (spatial) variability. (C) 2015 Elsevier Ltd. All rights reserved.