Recent advancements in information and communication technology have led to remarkable progress in the availability of urban spatial data. Among them, GIS data on population based on the national census is particularly well-developed. Urban planning practitioners, in particular, need to select the appropriate spatial unit（e.g., 250 meters, 500 meters, or 1 km）and criteria when examining changes in population density in a municipality. To this end, we develop a method for determining the appropriate grid square size for making histograms of population density based on computing the Akaike Information Criterion（AIC). Results of an empirical case study in Chiba prefecture show that histograms based on 1 km grid square is the best model of population density distribution in a municipality.
