Modifying current agricultural management practices as a means of sequestering carbon has been shown to be a relatively low cost way to offset greenhouse gas emissions. In this paper we examine the sensitivity of the estimates of the amount of soil carbon sequestered and the implied costs of sequestering a tonne of carbon to changes in the rates of soil carbon sequestered for alternative production practices. An application is made to the dryland grain production systems of the US Northern Plains where the marginal costs of soil C range from $20 to $100 per MT. We show that the resulting changes in the marginal costs quantities of C sequestered are not a monotonic transformation of the changes in the soil carbon rates. These results underscore the importance of using a linked economic and biophysical simulation model to assess the economic potential for sequestering carbon in agricultural soils.