The influence mechanism as well as impact path of built environment on public health was still hidden in the black box, and the existing western fragment analysis mostly failed to consider the issue of residential self-selection, which would cause estimation bias. To solve this problem, this research did a sample survey in the core urbanized area of Fuzhou so as to identify the mediating variables between urban built environment and the physiological health of residents. In order to avoid self-selection bias, robustness test was focused on the subgroups living in Danwei, Fanggaifang, or public housing communities that were unable to choose their own house independently. By using data of social survey, point of interest, road network and according to the structural equation model, functional density and diversity, road network accessibility, image maintenance and management, and accessibility of sports facilities had positive relationships with self-assessed physical health through the mediating role of physical activities. The above three characteristics had indirect positive effects on leisure walking behavior via accumulating community social capital, diminishing criminal activity and evaluating the sense of living security, while the moderate to vigorous physical activity was directly positively influenced by road network accessibility, functional density and diversity at the macro scale. Improving the accessibility of community sports facilities can promote both leisure walking and moderate to vigorous physical activity at the same time. Through mediating effect of healthy dietary behavior and unhealthy dietary behavior, the accessibility of healthy food was strongly negatively correlated with the number of chronic diseases. Even if the interference of residential self-selection was controlled within the subgroups, the above five types of built environment characteristics still had significant impact on physiological health, which proved that the optimization of built environment was indeed an effective and proactive way to improve public health. According to the standardized total effect, the influence of sports facilities and healthy food accessibility on physiological health was the highest, and the influence of the other three built environments was similar and moderate. Therefore, it was necessary to perfect both sports environment and food environment, while we evaluate both hardware environment (increasing road network density and functional diversity) and software environment (improving image maintenance and management) in healthy urban planning. The limitation of this paper is that the analysis was based on cross-sectional data and subjective measurements of physical activities and health indicators, which remained to be improved.