The incidence of upper limb and hand dysfunction after stroke is as high as 85%, severely impacting patients' quality of life. Traditional rehabilitation therapies have limited efficacy in some patients, making the exploration of novel rehabilitation techniques a research hotspot. Brain-computer interface (BCI), by decoding motor intention and providing real-time feedback, offers a new approach to inducing neuroplasticity and promoting motor function reconstruction. This article systematically reviews the clinical application progress, mechanisms of action, main paradigms, efficacy assessment, influencing factors, and future directions of BCI in upper limb motor rehabilitation after stroke. With the evolution of technology towards intelligence, the accumulation of high-level evidence, and the construction of a multimodal neural rehabilitation ecosystem, BCI is expected to grow from a cutting-edge technology into a reliable and transformative tool in neural rehabilitation therapy.