The study of "rare event" dynamics can be challenging despite continuing advances in computer hardware. A wide variety of methods based on the master equation approach have been developed to tackle such problems, where the focus is on Markovian dynamics between appropriately defined states. In this contribution, we employ the discrete path sampling approach to characterize pathways and rates for an adenine-adenine RNA conformational switch. The underlying free energy landscape supports competing structures separated by relatively high barriers, with the two principal funnels leading to the major and minor conformations identified by NMR experiments. The interconversion time scale is predicted to be a few hundred seconds, consistent with the experimental lower bound estimates. We find that conformational switching occurs via stacked intermediates, through a sliding mechanism, in agreement with a previous simulation study. By retaining full dimensionality and avoiding low-dimensional projections, the mechanism can be described at an atomistic level of detail.