Quantifying the relationship between vegetation phenology and urbanization level is crucial for exploring the impact of human activities on urban ecosystems. Here, we focused on the vegetation phenology of China's 35 typical cities and their surrounding areas. Based on the Normalized Difference Vegetation Index (NDVI) provided by the Moderate-resolution Imaging Spectroradiometer (MODIS), the Start of Growing Season (SOS), and the End of Growing Season (EOS) were extracted from the study area from 2000 to 2020 by Savitzky-Golay filter and dynamic thresholding method. The spatiotemporal variations of vegetation phenology in China's cities were analyzed in their urban-rural gradients, and the responses of phenology to Land Surface Temperature (LST) and Impervious Surface Area (ISA) in cities with different population sizes were compared as well. Results found that: (1) between 2000 and 2020, the SOS and EOS of urban vegetation generally exhibited a delaying trend with rates of 0.17 d/a and 0.15 d/a, while the SOS of megacities, megapolis, and small cities advanced particularly. The SOS of most southern cities and some cities in North China showed an advanced trend, and over 80% of the cities' EOS delayed. Overall, the SOS in the eastern and southern plains appeared earlier (average at DOY 91), while EOS appeared later (average at DOY 314), and SOS and EOS in the western and northern urban regions expressed an opposite trend. (2) In 63% of the analyzed cities, SOS advanced as the distance from the urban area shortened (0.6-4.3 d/km), while in 60% of these cities, EOS delayed (0.2-1.9 d/km) as the distance to the urban area decreased. (3) The spring LST of all cities presented a consistent increasing trend. For every 1℃ increase of ΔLST in spring, the SOS advanced 6.8 days. For every 1℃ increase in autumn ΔLST, EOS delayed 1.5 days. There was a significantly negative correlation between urban vegetation SOS and ISA proportion, with a 1% increase in ISA proportion causing a 0.253-day advance in SOS. Additionally, there was a significantly positive correlation between the vegetation EOS and ISA proportion, and when the proportion of ISA increased by 1%, EOS delayed by 0.106 days. Furthermore, there were differences in the response of vegetation phenology to LST and ISA in cities with different population sizes. The influence of LST and ISA in the megapolis and big cities was higher than that in the other three size cities, indicating that the vegetation phenology in the megapolis and big cities was more sensitive to LST and ISA in the urban-rural gradient.