Computer simulation approaches in biomolecular recognition processes have come a long way. In this Perspective, we highlight a series of recent success stories in which computer simulations have played a remarkable role in elucidating the atomic resolution mechanism of kinetic processes of protein-ligand binding in a quantitative fashion. In particular, we show that a robust combination of unbiased simulation, harnessed by a high-fidelity computing environment, and Markov state modeling approaches has been instrumental in revealing novel protein-ligand recognition pathways in multiple systems. We also elucidate the role of recent developments in enhanced sampling approaches in providing the much-needed impetus in accelerating simulation of the ligand recognition process. We identify multiple key issues, including force fields and the sampling bottleneck, which are currently preventing the field from achieving quantitative reconstruction of experimental measurements. Finally, we suggest a possible way forward via adoption of multiscale approaches and coarse-grained simulations as next steps toward efficient elucidation of ligand binding kinetics.