Composite plate shear walls with concrete filled (C-PSW/CF) are widely used in industrial and high-rise buildings due to construction convenience and excellent load-bearing performance. However, buckling of the steel plate is an important factor that reduce the in-plane bearing capacity of C-PSW/CF, and thus C-PSW/CF’s stiffness degrades rapidly which eventually cause failure or collapse of shear walls. The buckling behavior is difficult to be considered in C-PSW/CF. In this paper, a column-supported rotational-constraint (CSRC) plate model was proposed based on plate theory and numerical simulations. The slenderness limit calculated by this model showed good agreement with design codes, indicating the validation of CSRC model. Then, uncertainties from geometry, material and boundary condition were separately introduced, and thousands of finite element models based on CSRC plate model were generated with stochastic harmonica function (SHF) method, which contained the statistic information of one of these three types of uncertainties. These models were analyzed by using eigenvalue method and the results showed that geometric imperfections can cause a 12% degradation of mean critical buckling loads when the correlation length is small while almost no decrease observed with large correlation lengths at most cases. And the uncertainties from material and boundary condition hardly affects the mean critical buckling loads. The simulation results illustrated that morphology fluctuation of steel plate surface with small correlation lengths can lead to a significant reduction in the buckling resistance of steel plates. Furthermore, an engineering case of corrosion on steel plates under urban atmosphere was analyzed with eigenvalue method, in which translation theory was used to generate cumulative lognormal distribution of corrosion depth with a certain auto-correlation function of the Gaussian type. The results showed that the mean and 95th percentile of the critical buckling loads decreased by about 10% when the exposure duration was about 2.5 years, and the mean and 95th percentile of the critical buckling loads decreased dramatically to 50% of the control value when the exposure duration was more than 10 years. It is suggested that the buckling resistance of corroded shear walls need to be taken into accounted.