Human food systems are a key contributor to climate change and other environmental concerns. While the environmental impacts of diets have been evaluated at the aggregate level, few studies, and none for the US, have focused on individual self-selected diets. Such work is essential for estimating a distribution of impacts, which, in turn, is key to recommending policies for driving consumer demand towards lower environmental impacts. To estimate the impact of US dietary choices on greenhouse gas emissions (GHGE) and energy demand, we built a food impacts database from an exhaustive review of food life cycle assessment (LCA) studies and linked it to over 6000 as-consumed foods and dishes from 1 day dietary recall data on adults (N = 16 800) in the nationally representative 2005–2010 National Health and Nutrition Examination Survey. Food production impacts of US self-selected diets averaged 4.7 kg CO2 eq. person−1 day−1 (95% CI: 4.6–4.8) and 25.2 MJ non-renewable energy demand person−1 day−1 (95% CI: 24.6–25.8). As has been observed previously, meats and dairy contribute the most to GHGE and energy demand of US diets; however, beverages also emerge in this study as a notable contributor. Although linking impacts to diets required the use of many substitutions for foods with no available LCA studies, such proxy substitutions accounted for only 3% of diet-level GHGE. Variability across LCA studies introduced a ±19% range on the mean diet GHGE, but much of this variability is expected to be due to differences in food production locations and practices that can not currently be traced to individual dietary choices. When ranked by GHGE, diets from the top quintile accounted for 7.9 times the GHGE as those from the bottom quintile of diets. Our analyses highlight the importance of utilizing individual dietary behaviors rather than just population means when considering diet shift scenarios.