Buildings are the largest contributor to energy demand, greenhouse gas emissions, resource consumption, and waste generation. To assess the environmental impacts (such as resource use and emissions) of a product throughout its entire life cycle, life cycle assessment (LCA) has been introduced as a powerful tool. On the other hand, machine learning has been widely explored and applied in buildings research over the past decades, demonstrating its potential to enhance building performance. Machine learning with LCA methods may contribute greatly to lowering the environmental impacts. However, the multitude number of input parameters and uncertainties for life cycle analysis make it difficult to comprehend the use and capabilities of machine learning in LCA. This research conducted a critical review on previous research and application of machine learning in LCA to examine how machine learning has been used in buildings’ LCA. The results are presented in terms of three levels of LCA in buildings: LCA of building materials and components, LCA of individual buildings, and LCA of building sector. The results reveal that machine learning techniques were effective in certain aspects of LCA. This research also discusses the limitations and gaps identified and proposes future direction to advance future application of machine learning for LCA in the built environment.