Abstract: As the urbanization process has entered the “second half”, the climate-responsive urban design based on the green sustainable theory has attracted increasing attention. The urban form, as the core of urban design, has also been brought into the focus. However, existing studies lack detailed urban form data and ignore the effect of urban morphological characteristics at the human scale. In this context, this research conducts a large-scale and high-precision study in the central urban area of Beijing. It applies the Landsat 8 remote sensing image data to calculate street temperatures based on the radiative transfer equation method. It also integrates multisource data, such as streetscape data, building form data and point of interest data, to build a multi-level urban form regression model to quantify the effects of top-down morphological perspective and bottom-up human scale on street temperatures. Statistical analysis shows that from the top-down morphological perspective, street spatial location, surrounding building density, floor area ratio, functional diversity, and distance between blue and green space have significant impacts on street temperatures. With regard to the bottom-up human scale, it shows that the higher the proportion of street green visibility and building interface, the better shading effect. These findings can help improve the efficiency of specific street design, make differentiated street design strategies according to street attributes, and facilitate the deepening of climateresponsive urban design from the perspective of urban morphology.