In order to analyze competing policy approaches for addressing global climate change, a wide variety of economic-energy models are used to project future carbon emissions under various policy scenarios. Due to uncertainties about future economic growth and technological development, there is a great deal of uncertainty in emissions projections. This paper demonstrates the use of the Deterministic Equivalent Modeling Method, an efficient means for propagating uncertainty through large models, to investigate the probability distributions of carbon emissions from the MIT Emissions Prediction and Policy Analysis model. From the specific results of the uncertainty analysis, several conclusions with implications for climate policy are given, including the existence of a wider range of possible outcomes than suggested by differences between models, the fact that a "global emissions path through time" does not actually exist, and that the uncertainty in costs and effects of carbon reduction policies differ across regions.