All-atom molecular dynamics (MD) simulations provide a powerful way to guide our thinking about links between molecular structure and macroscopic properties, but only if we continue to explore our understanding of, and overcome, its limitations. The most important limitation of MD is arguably the a**rare-eventa** problem: transitions between recognizable metastable states occur too infrequently for MD simulation to statistically characterize. In this talk, Ia**ll highlight two recent methods developed in our group to address the rare-event problem in MD. The first, Markovian Milestoning, permits computation of rate constants associated with relatively slow processes in all-atom systems. We show its application to the problem of structurally understanding the influence of substrate binding on the entry and exit kinetics of oxygen in flavoenzymes. The second, replica-exchange on-the-fly free-energy parameterization, overcomes the classical a**hidden-variablea** problem prevalent in rare-event methods in biomolecular simulations. We show its application in the understanding of the effects of phenotypically distinct mutations in the human prion protein, the etiological agent of Creutzfeldt-Jakob disease and other neurodegenerative diseases. The talk will end with some perspectives on the future of rare-event methods in furthering our structural understanding of biomolecular kinetics.