Abstract: The light environment created by natural light is preferred by public due to its own economic, natural, pleasant and irreplaceable characteristics. Natural lighting is not only conducive to energy conservation of lighting, but also conducive to increasing the exchange of natural information indoor and outdoor, improving the space health environment and regulating the mood of space users. Making full use of the natural light in the building is of great significance for creating a good light environment, saving energy, protecting the environment and building green buildings. Therefore, it is necessary to optimize the lighting design of buildings. This paper puts forward an idea of natural lighting analysis in building based on the BIM technology and the BP neural network. Taking the CDUT library (library of Chengdu University of Technology) as an example, a 3D visualization model is established by using Revit software to generate the building information file in gbXML format. Then, the gbXML file is imported into Ecotect software to simulate and analyze the indoor light environment of the library, to calculate the natural lighting coefficient, and to quantitatively analyze the impact of window height, glass transmittance and wall material light reflectivity on the indoor light environment. At last, with the help of Weka software, the neural network model based on BP algorithm is established, obtaining the BP neural network model which can predict the variation range of variables under the optimal lighting coefficient.