This chapter introduces the preprocessing methods for blind source separation in detail. Firstly, various whitening algorithms are investigated, which can reduce the dimensionality of mixed signal matrix and complexity of the problem. Among these algorithms, robust whitening algorithms can eliminate the influence of noise to a certain extent, and therefore they are more commonly used than standard whitening algorithms. Then three types of existing source signal number estimation methods are studied, the algorithm principle and implementation process are derived in detail, and finally, the performance of the source signal number estimation methods under different conditions is evaluated in simulation. From a comprehensive point of view, the methods based on the MDL criterion and the methods based on eigenvalue decomposition have better estimation performance and can estimate the number of source signals more accurately.