Existing methods for large regional population density estimation, which are mostly concentrated in the kilometer scale and only reflect the macro distribution characteristics of the urban and rural population, are difficult to describe details of urban and rural population spatial distribution accurately. In order to resolve the problem above, an estimation method of large regional urban and rural population density, which is based on the first 30 m global land cover dataset(GlobeLand30) is proposed. Based on the urban and rural area data partitioned from artificial surfaces data in GlobeLand30 datasets, the population density were estimated in urban and rural area respectively. Urban population density was estimated through the correlation between night lighting intensity and population. Through area revise of rural patches by the method of quadrats estimation, the rural population density was estimated. This paper takes Shandong province as a test area. The result shows that the method of urban-rural population density estimation could reflect the heterogeneity and continuity of the population spatial distribution in urban internal well, and express the population spatial distribution in rural area. By comparison with the reference data, the method of this paper is superior to the reference data in describing the spatial extent of residents and expressing the spatial distribution of population. And due to the globality of GlobeLand30 data, it is feasible to extend the method to a wider area.