The transport of runaway electrons in chaotic magnetic geometries is an area of active study.The transport of runaway electrons in chaotic magnetic geometries is an area of active study. Computing the transport from the direct simulation of particle trajectories is computationally expensive. Instead, diffusion models, such as the one by Rechester and Rosenbluth (RR), are often employed to incorporate transport effects into reduced simulations. However, the comparison of diffusion-based to direct simulations reveals that the transport is typically not purely diffusive. In this paper, we introduce a simple transport model, based on chaos theory, which goes beyond the RR approximation. Besides chaotic diffusion, our model takes into account the effect of socalled sticky regions, a trapping layer around magnetic islands, where particle escape slows down to a power-law decay rather than an exponential decay. We demonstrate the applicability of the model both in the Ullmann-Caldas map with parameters corresponding to the TBR-1 tokamak, and in a JOREK simulation of a Joint European Torus disruption scenario, with remarkably good fits achieved in both cases. Computing the transport from the direct simulation of particle trajectories is computationally expensive. Instead, diffusion models, such as the one by Rechester and Rosenbluth (RR), are often employed to incorporate transport effects into reduced simulations. However, the comparison of diffusion-based to direct simulations reveals that the transport is typically not purely diffusive. In this paper, we introduce a simple transport model, based on chaos theory, which goes beyond the RR approximation. Besides chaotic diffusion, our model takes into account the effect of socalled sticky regions, a trapping layer around magnetic islands, where particle escape slows down to a power-law decay rather than an exponential decay. We demonstrate the applicability of the model both in the Ullmann-Caldas map with parameters corresponding to the TBR-1 tokamak, and in a JOREK simulation of a Joint European Torus disruption scenario, with remarkably good fits achieved in both cases.