Building integrated photovoltaics is an important measure to promote low-carbon urban growth. However, the solar utilization performance of buildings in a block is greatly influenced by the shadings from surrounding buildings with di-verse layouts and morphologies. Therefore, this study evaluates the effect of urban morphology on solar energy potential for buildings in diverse urban environments using the parametric modelling and deep learning approaches. By con-trolling urban morphology parameters in different ranges, thousands of block models are randomly generated for residential buildings based on a parametric approach. Then, the solar energy potential, including the solar radiation and photovoltaics power potential, can be evaluated for building roofs and facades. The contribution of morphological parameters to building solar energy potential are prioritized using Global Sensitivity Analysis. The different input combinations of morphology parameters can be obtained based on the total contribution of each parameter. These input combinations are inserted into deep learning models to explore different prediction performances on the target values. Correspondingly, the influential parameters that can well predict the solar performance of building roofs and facades are selected. The proposed approach and findings are expected to offer inspirations for solar design and utilization in urban buildings.