We quantify how changes in natural flood discharge control bedload rest time distributions and may influence particle diffusion through mountain river networks. We embedded accelerometers and gyroscopes into artificial cobbles deployed in Halfmoon Creek, Colorado, USA, and measured bedload transport during 28 daily snowmelt flood hydrographs in 2015. From the motion sensor data we calculate motion and rest distributions over ~6 orders of temporal magnitude, from ~2 seconds to ~1 month. Motion durations follow a thin-tailed exponential distribution. Rests >12 hours can be well fit by both truncated Pareto distributions and exponentially-tempered Pareto distributions, suggesting ambiguity in whether rests remain heavy-tailed or transition to thin tails at even longer timescales. Rest time scaling varies not only with timescale but also with flow intensity, becoming less heavy-tailed as shear stress increases. A rest time scaling break at ~12 hours may be caused by daily discharge cyclicity.