Spatial pattern of population distribution has a typical character of scale dependency. Fine-scale estimation of the population distribution has been a huge challenge in the field of population geography. In China, township is the finest administrative unit in official population census data. Thus, population density data at township level can be used to describe and characterize the population spatial pattern and changes elaborately, and support optimized layout plan of Chinese population, and government policy decision making. Township boundaries across 25 provinces in China had been collected in this study. The 5th national population census data was spatially joined to the boundary layer for population density calculation. Methods of Lorenz curve, geo-spatial analysis, and latitude/longitude transect had been applied to reveal the agglomeration degree, spatial patterns of population distribution with latitude and longitude. Based on the correlation and stepwise regression analysis, four variables, including relief degree of land surface (RDLS), water system density (WSD), road system density (RSD) and nighttime light index (NTL), were introduced to check the variety of population distribution in each province. The results showed that: (1) the variety of population distribution can be distinguished clearly by the mean population density on township scale than that of county level. (2) The overall population distribution can be characterized as dense in the southeast and sparse in the northwest, while lower population density occurs in some part of southeastern China, and vice versa. (3) The population density at township level along six designed longitude and latitude transects varies greatly. The possible impact factors include the three-level landscape features, local topography, regional economy, and the proximity to coastline, transportation hub, and hydrological systems. The correlation coefficient at provincial level, between the RDLS, WSD, RSD, NTL, and the population density has been identified as 0.56, 0.28, 0.61, and 0.69, respectively. At township level, topography and economy exert more impacts on population distribution in Liaoning, Jilin, Beijing, Tianjin, Shanghai, Hebei, Henan, Shaanxi, Shanxi, Shandong, Anhui, Jiangsu, Hunan, Hubei, Jiangxi, Zhejiang, Fujian, Guangdong, and Hainan. In addition to the four factors mentioned above, it is necessary to introduce more natural or social factors to explore the population distribution pattern in Tibet, Qinghai, Inner Mongolia, Yunnan, and Guizhou. This study expanded the research scale and dimension of the research in population geography research in China. The resulted population density dataset in 2000 is expected to enrich the baseline data resources for population geography development in China and the world.