Histopathological diagnosis is considered the “gold standard” for clinical cancer diagnosis. The deep learning-based computer-aided diagnosis (CAD) for histopathological images can help pathologists improve diagnostic accuracy and efficiency. However, there are differences in the stained histopathological images among different hospitals, which can subsequently affect the generalization performance of CAD models trained only with samples from one hospital. Although multi-center learning can address this issue, it remains challenging to gather and share a large number of histopathological images from different hospitals for model training due to various factors, such as privacy protection and data security. Federated learning is a distributed machine learning method, which shares the local model parameters across different centers instead of the local data of different centers in the traditional multi-center learning paradigm. Therefore, federated learning can effectively address the aforementioned issue. Currently, there are studies on federated learning-based CAD for histopathological images, which are surveyed in this paper. It first introduces deep learning-based CAD for histopathological images, then presents the content of federated learning and the federated learning-based CAD for histopathological images. Finally, the conclusion and future work are given.