Abstract: In order to investigate the evolution patterns of city size in China, nighttime light images were used to analyze city size distribution and its spatiotemporal patterns for 338 prefectural cities from 1993 to 2012 based on total nighttime light index, rank-size analysis and Markov transition matrix analysis. Statistical analysis shows that, all provincial regions and economic zones have experienced nighttime light growth, with a growth rate of 257% in the Southwest and 254% in the Northwest. Tibet is the fastest growing provincial region with a growth rate of 842%. The rank-size analysis shows all the other economic zones, except for the Northeast economic zone, become less agglomerated during this period. The Markov transition matrix analysis shows the city size has inertia from 1993 to 2012. Cities in lowest level remain their levels with a probability of 82%. In addition, the size of a city may transit to the neighboring level, but the crossing-level transition rarely happens. This study reveals the city size evolution patterns in a multiple spatial scale and a long period, providing evidences to evaluate policies for Chinese urban development.