Abstract: With the mechanical energy stored in the rotor, wind turbine generators can provide inertia response to support frequency stability. However, the unreasonable settings of virtual inertia parameters may threaten the small-signal stability of power systems. Thus, this paper proposes a virtual inertia parameter optimization method that can satisfy system probabilistic stability. Considering the uncertainty of wind power, the virtual inertia parameters optimization model is established to achieve the maximum probability of satisfying frequency stability and small-signal stability. For the difficulty of solving the probabilistic optimization model, the frequency probabilistic stability objective is transformed into a determined frequency stability constraint by considering the extreme fault of the system. Besides, the analytical expression of small-signal probabilistic stability is established by the analytical cumulative method. Taking the damping ratio as the intermediate variable, the sensitivity of objective function and virtual inertia is solved. The Newton method with sensitivity as a gradient is used to solve the optimization model. Case studies demonstrate the proposed model can improve probabilistic small-signal stability while ensuring frequency stability.