This PhD thesis focuses on improving reliability of the LCA of buildings by a better control of the origin and magnitude of uncertainties. It contributes to: - The development of a methodology for assessing uncertainty in LCA of buildings; - The construction and enrichment of a database capitalizing uncertainties at material and product scales, which would be useful in the assessment of building impacts’ uncertainties; - The development of a contribution and sensitivity analysis methodology, which identifies the key parameters influencing the robustness of results. - The development of a methodology for robust comparisons among projects based on contribution and sensitivity analysis. - The identification of optimal environmental dimensions of building structures. This study has also identified the evolution of environmental impacts based on the parameters of structural design. For optimum dimensions the environmental impacts of two reinforced concrete structures were performed. The methodologies used in this research are suitable for the developers of LCA software to enhance the functionality of making comparisons among projects (this work can be implemented in the ELODIE software). The results can be used by architects, civil engineers, thermal engineers, statisticians, environment and life cycle engineers, allowing them to identify the most important contributors to the environmental impacts of buildings and their uncertainties.