<p>Reconstructing paleoclimate characteristics and understanding their evolution rules is a key question in Earth system science and global change research. It helps clarify the historical position of the modern warming period, understand the features and mechanisms of climate change under warming backgrounds, and thereby improve the accuracy of future climate projections. Proxy records and numerical modeling are two primary approaches in current paleoclimate study. As an emerging methodology, paleoclimate data assimilation effectively integrates paleoclimate proxy records with numerical simulations, combining their advantages to enhance the accuracy of paleoclimate reconstructions. This paper systematically reviews recent progresses in paleoclimate data assimilation. It first outlines the historical developments of major paleoclimate data assimilation methods, discusses their advantages, disadvantages, and applicability, and highlights recent improvements such as the application of machine learning methods and the developments of online assimilation. Then, it introduces applications of paleoclimate data assimilation according to different typical paleoclimate periods, particularly addressing challenges associated with various types of proxy data, and summarizes currently available open-source datasets and algorithm platforms. A specific case study is presented to illustrate the application of assimilating oxygen isotope simulations. Finally, the paper discusses unresolved issues and challenges in paleoclimate data assimilation studies, and outlines potential directions for future research.</p>