This paper, based on China's 30 provinces' panel data from 1999—2010, estimates their carbon emission quantity per year, and applies static and dynamic panel data models to study through cases the factors of carbon emission. The results show a topdown U curve between carbon per capita and carbon emission per capita. The major factors are composed of energy consumption structure, urbanization level and industrial structure. The current carbon dioxide emission per capita rises with the last emission. Intensifying research & development and trade openness can suppress carbon dioxide emission.