In recent years, millimeter-wave radar technology has been widely used for non-invasive recognition and tracking of sleep postures due to its advantages of high accuracy, contactless operation, and ability to penetrate clothing. In order to promote the development of this field and to address the lack of large-scale, high-quality sleep posture transition datasets, this paper proposes a publicly available millimeter-wave sleep posture transition dataset. The dataset contains 20 volunteers (15 males and 5 females) aged between 19 and 25 years, with heights ranging from 1.55 m to 1.80 m and weights between 45 kg and 90 kg. Each participant performed seven different body position transitionmaneuvers in a preset order, yielding a total of 1400 samples. During the experiment, participants' postural changes were captured by a millimeter-wave radar system mounted on the side of the bed. This dataset provides valuable support for the optimization of sleep posture recognition algorithms, analysis of nocturnal behavioral patterns, and health monitoring.