To demonstrate and support sustainability of large-scale infrastructure projects, life cycle assessment (LCA) has become a common tool used in the evaluation of environmental impact. However, LCA includes a number of phases, which are all associated with uncertainties. Decisions made without regard to these uncertainties may be suboptimal and/or misleading. The current paper discusses some important issues about how to deal with uncertainties in LCA. Two case studies are presented that shows the effect of including uncertainty in various parameters. The results show that a reduction in uncertainty due to manufacturing improvements can increase the reliability of the results significantly. It was also found that the uncertainty assumptions for parameters where data is not available have a large impact on the LCA-results.