This paper employs analytical and numerical general equilibrium models to assess the efficiency impacts of two policies to reduce U.S. carbon emissions - a revenue-neutral carbon tax and a non-auctioned carbon quota - taking into account the interactions between these policies and pre-existing tax distortions in factor markets. We show that tax interactions significantly raise the costs of both policies relative to what they would be in a first-best setting. In addition, we show that these interactions put the carbon quota at a significant efficiency disadvantage relative to the carbon tax: for example, the costs of reducing emissions by 10 percent are more than three times as high under the carbon quota as under the carbon tax. This disadvantage reflects the inability of the quota policy to generate revenue that can be used to reduce pre-existing distortionary taxes. Indeed, second-best considerations can limit the potential of a carbon quota to generate overall efficiency gains. Under our central values for parameters, a non-auctioned carbon quota (or set of grandfathered carbon emissions permits) cannot increase efficiency unless the marginal benefits from avoided future climate change are at least $17.8 per ton of carbon abatement. Most estimates of marginal environmental benefits are below this level. Thus, our analysis suggests that any carbon abatement by way of a non-auctioned quota will reduce efficiency. In contrast, our analysis indicates that a revenue-neutral carbon tax can be efficiency-improving so long as marginal environmental benefits are positive.