Non-invasive continuous blood pressure monitoring is a core technology for the prevention, diagnosis, and management of cardiovascular diseases. Traditional cuff-based measurement methods have limitations such as intermittent readings, patient discomfort, and the “white coat effect”. With the rapid advancement of wearable technology and mobile healthcare, non-invasive continuous blood pressure measurement based on multi-modal physiological signal fusion has emerged as a research hotspot. Research progress in multi-modal physiological signal fusion for non-invasive continuous blood pressure measurement encompasses the characteristics and blood pressure correlations of commonly used physiological signals, as well as multi-modal signal acquisition techniques and sensor integration solutions. This paper provides an in-depth analysis of data-level, feature-level, and decision-level approaches, as well as end-to-end fusion methods based on deep learning. It compares the performance and applicability of different fusion strategies, summarizes the challenges currently faced by wearable device applications, and outlines future development directions. Research indicates that multi-modal fusion can effectively integrate complementary information from various signals, significantly improving the accuracy and robustness of blood pressure estimation. However, individual variability, dynamic interference, data standardization, and model interpretability remain core challenges. Future trends will focus on the establishment of high-quality clinical datasets, the development of novel wearable sensing technologies, intelligent multimodal signal processing approaches, and personalized adaptive blood pressure estimation models.These efforts will drive clinical translation and widespread adoption of this technology, providing robust technical support for achieving early hypertension warning and precision management.