Several kinds of policies have been advanced to take advantage of agriculture’s climate change mitigation potential. Under cap-and-trade programs (such as California’s AB32), we allow greenhouse gas emissions from fossil fuel combustion in regulated sectors such as power generation in exchange for increased adoption of agricultural management practices believed to have climate benefits (“offsets”). A second approach incentivizes the adoption of “climate-friendly” agriculture using public funds but without affecting regulation of industries in other sectors. A third relies strictly on voluntary measures, providing farmers with tools to estimate the influence of alternative management practices on GHG emissions. These three approaches all rely – but to highly varying extent -- on our ability to quantify and account for changes in greenhouse gas emissions from changes in agricultural practices. How well can we quantify these reductions, and what are the consequences if our estimates are wrong? Our ability to estimate and control the risk of project failure depends partially upon the type of mitigation proposed. Projects intended to enhance soil carbon storage face a higher risk of reversal than projects aimed at decreasing nitrous oxide emissions, but the actual risk of soil carbon release is hard to quantify. Taking into account issues of quantifying GHG emissions associated with changing agricultural practices and issues of risk, I will present a framework for matching policy tools with agricultural practices intended to mitigate climate change. I will use this framework to recommend types of mitigation that should receive priority under existing policies, and features that future policies should include.