Aberrant activation of Ras GTPase is closely associated with human cancers. Mechanistic and structural insights into conformational transitions underlying the Ras deactivation pathway provide opportunities for the development of targeted therapies for the treatment of Ras-driven cancers. It remains challenging, however, to completely capture a large-scale conformational transition by atomistic molecular dynamics (MD) simulations. Here, we carry out a computational scheme that combines a transition pathway generation algorithm, extensive MD simulations, and Markov state model analysis for disclosing the conformational landscape of the Ras deactivation pathway. Our findings suggest a stepwise deactivation pathway for Ras hydrolysis and identify several key conformational substates along the deactivation pathway. Furthermore, we discover an unexplored and potentially allosteric binding site on the effector-binding region of Ras in the conformational substates, which is further supported by site-directed mutagenesis experiments. This site could be exploited to the design of Ras inhibitors to block Ras–effector interactions.