Real-time population data are vital for urban planning and resource management for sustainable development. To complement satellite-based population estimation methods, geospatial social media data provide additional opportunities to estimate the distribution of population with high levels of efficacy and accuracy. Thus, this study attempts to assess the performance of various sensing data to disaggregate population data in China; the tested data include Tencent location-based service (LBS) data (about 0.8 billion users), satellite-derived land use/cover data, and nightlight imagery data. With the use of census data for validation, the experimental results show that Tencent LBS data are much better than satellite-derived land use/cover data and nightlight satellite data for mapping the population distribution. The overall mapping accuracy at the city level using Tencent LBS data was 88.9%, whereas the accuracy using land use/cover data was 87.1% and that using nightlight satellite data was 85.5%. The experimental results also indicate that LBS data and remote sensing data could both be well integrated to map the population distribution in China. Thus, a population spatialization model was further developed using all of the tested indicators; this model allowed the overall population estimation accuracy at the city level to reach 90.4%. This model could help determine the population distribution on various spatial scales quickly and efficiently, and the developed tool and the provided population estimates may be vital for the sustainable development of cities and regions for which population data are lacking.