Abstract: This paper presents an urban population sensing method based on mobile phone dataset and applies it to San Francisco Bay Area. The difference of mobile phone market shares in different tracts is considered. We introduce the potential application of dynamic population sensing using mobile phone data and calculate the relative difference of daytime population and nighttime population in different tracts in Bay Area. The knowledge of urban population distribution has great importance of the early-warning of city emergency, urban traffic control, and the allocation of city public resources.