By using a stochastic simulation model based on the kinetic Monte Carlo approach, we study the physics, operation and reliability of resistive random-access memory (RRAM) devices based on oxides, including silicon-rich silica (SiO) and hafnium oxide – HfO – a widely used transition metal oxide. The interest in RRAM technology has been increasing steadily in the last ten years, as it is widely viewed as the next generation of non-volatile memory devices. The simulation procedure describes self-consistently electronic charge and thermal transport effects in the three-dimensional (3D) space, allowing the study of the dynamics of conductive filaments responsible for switching. We focus on the study of the reliability of these devices, by specifically looking into how oxygen deficiency in the system affects the switching efficiency. © 2020, Springer Nature Switzerland AG.