Recently, Fourier transform has been widely introduced into deep neural networks to further advance the state-of-the-art regarding both accuracy and efficiency of time series analysis. The advantages of the Fourier transform for time series analysis, such as efficiency and global view, have been rapidly explored and exploited, exhibiting a promising deep learning paradigm for time series analysis. However, although increasing attention has been attracted and research is flourishing in this emerging area, there lacks a systematic review of the variety of existing studies in the area. To this end, in this paper, we provide a comprehensive review of studies on neural time series analysis with Fourier transform. We aim to systematically investigate and summarize the latest research progress. Accordingly, we propose a novel taxonomy to categorize existing neural time series analysis methods from four perspectives, including characteristics, usage paradigms, network design, and applications. We also share some new research directions in this vibrant area.