If carbon sequestration is to be a cost-effective substitute for reducing emissions then it must occur under a framework that ensures that the sequestration is additional to what would otherwise have occurred, the carbon is stored permanently, and any leakage is properly accounted for. We discuss significant challenges in meeting these requirements, including some not previously recognized. Although we focus on sequestration in soil, many of the issues covered are applicable to all types of sequestration. The common-practice method for determining additionality achieves its intention of reducing transaction costs in the short term but not in the medium to long term. Its design results in the least costly, additional abatement-measures being excluded from policy support and fails to address how, in the case of sequestration, revisions to the additionality of sequestering practices should apply not just to the future, but in theory, also retrospectively. Permanence is sometimes approximated as 100 years of sequestration. Re-release of sequestered carbon after this will not only reverse the sequestration, but may raise atmospheric carbon to higher levels than they would have been if the sequestration had never occurred. Leakage associated with sequestration practices can accumulate over time to exceed the total level of sequestration; nonetheless, adoption of such practices can be attractive to landholders, even when they are required to pay for this leakage at contemporary prices. Policy relevance Globally, much has been written and claimed about the ability to offset emissions with sequestration. The Australian Government plans to use sequestration to source much of the abatement required to reach its emissions targets. Designing effective policy for sequestration will be challenging politically, and will involve substantial transaction costs. Compromises in policy design intended to make sequestration attractive and reduce transaction costs can render it highly inefficient as a policy.