Policies formulated to reduce greenhouse gas (GHG) emissions, such as a low-carbon fuel standard, frequently rely on life-cycle assessment (LCA) to estimate emissions, but LCA results are often highly uncertain. This study develops life-cycle models that quantitatively and qualitatively describe the uncertainty and variability in GHG emissions for both fossil fuels and ethanol and examines mechanisms to reduce those uncertainties in the policy process. Uncertainty regarding emissions from gasoline is non-negligible, with an estimated 90% confidence interval ranging from 84 to 100 g CO(2)e/MJ. Emissions from biofuels have greater uncertainty. The widths of the 90% confidence intervals for corn and switchgrass ethanol are estimated to be on the order of 100 g CO(2)e/MJ, and removing emissions from indirect land use change still leaves significant remaining uncertainty. Though an opt-in policy mechanism can reduce some uncertainty by incentivizing producers to self-report fuel production parameters, some important parameters, such as land use change emissions and nitrogen volatilization, cannot be accurately measured and self-reported. Low-carbon fuel policies should explicitly acknowledge, quantify, and incorporate uncertainty in life cycle emissions in order to more effectively achieve emissions reductions. Two complementary ways to incorporate this uncertainty in low carbon fuel policy design are presented. (C) 2012 Elsevier Ltd. All rights reserved.