The intrinsic variability of the switching parameters in resistive memories has been a major wall that limits their adoption as the next generation memories. In contrast, this natural stochasticity can be beneficial for other applications such as Random Number Generators (RNGs). This paper presents two RNG approaches based on a 130nm HfO2-based Resistive RAM (RRAM) memory array. The memory array is programmed with a voltage close to the median value of the SET (resp. RESET) voltage distribution to benefit from the SET (resp. RESET) voltage variability. In both cases, only a subset of the memory array is programmed, resulting in a stochastic distribution of cell resistance values. Resistance values are next converted into a bit stream and confronted to National Institute of Standards and Technology (NIST) test benchmarks.