A disc-pad system is established to study impacts of surface topography on brake squeal from the perspective of statistical analysis. Firstly, surface topographies of brake disc and pad are precisely measured on the scale of micron and are statistically analyzed with a three-dimensional evaluation system. Secondly, the finite element model of brake disc and pad without surface topographies is created and verified through component free modal tests. Thereby the valid brake squeal model for complex modal analysis is built with ABAQUS. An effective method is developed to apply interface topographies to the smooth contact model, which consequently establishes sixty brake squeal models with topographies. Thirdly, impacts of surface topography on brake squeal are studied through comparison and statistical analysis of prediction results with and without topographies.The analysis manifest that topography amplitudes and evaluation index deviations of brake pad far exceed those of the disc, indicating the surface of brake pad is relatively much rougher. Moreover, squeal prediction results confirm that surface topographies can both cause contact pressure variation and affect the randomness of brake squeal characteristics, namely, numbers and values of squeal frequencies. Additionally, contact interface topographies can cause system modal coupling states to change, which further affects squeal tendency and uncertainty. Meanwhile, statistical characteristics and variations of surface topographies demonstrate close correlation with those of brake squeals. Hence the consideration of surface topographies can effectively improve squeal prediction precision, and stochastic characteristics of brake squeal during tests can be well justified by the time-varying surface topography.