Abstract: Understanding the dynamics of population distributions at both fine spatial and temporal scales is crucial for urban planning and disaster risk management. Traditionally, accurately mapping the real-time spatial distribution of a population has posed challenges because of limitations in the available data. The extensive and continuous sampling of mobile location-based data now provides new possibilities for estimating population distributions with a high level of spatiotemporal resolution. In this study, a real-time population distribution estimation model was developed utilizing both mobile phone user-location signal (MPLS) data and population census data. The real-time variations in population distribution were assessed at fine spatial and temporal scales within Xining City, China, based on MPLS data. The findings reveal that MPLS data provide detailed insights into the spatial and temporal variations in population distribution within the city. A highly significant correlation exists between the MPLS-based population estimations and the census figures, with a correlation coefficient of 0.94 at a significance level of 0.001. While densely populated areas may maintain a relatively concentrated spatial distribution throughout the day, there are noteworthy fine-scale spatiotemporal variations in population distribution attributed to the daily activities of individuals in Xining City. At the town scale, the overall change rate of population distribution at different times during a day ranged from −50% to 43%. The finer the estimation scale is, the more significant the changes in real-time population distribution. Real-time population distribution estimation can provide a more objective and accurate reference for local disaster risk management. The integration of multisource location-based data and the continuous improvement of data mining techniques are likely to further increase the accuracy of real-time population distribution estimations, which holds significant potential for reducing disaster risk.