Carbon burial is increasingly valued as a service provided by threatened vegetated coastal habitats. Similarly, shellfish reefs contain significant pools of carbon and are globally endangered, yet considerable uncertainty remains regarding shellfish reefs' role as sources (+) or sinks (−) of atmospheric CO2. While CO2 release is a by-product of carbonate shell production (then burial), shellfish also facilitate atmospheric-CO2 drawdown via filtration and rapid biodeposition of carbon-fixing primary producers. We provide a framework to account for the dual burial of inorganic and organic carbon, and demonstrate that decade-old experimental reefs on intertidal sandflats were net sources of CO2 (7.1 ± 1.2 MgC ha−1 yr−1 (µ ± s.e.)) resulting from predominantly carbonate deposition, whereas shallow subtidal reefs (−1.0 ± 0.4 MgC ha−1 yr−1) and saltmarsh-fringing reefs (−1.3 ± 0.4 MgC ha−1 yr−1) were dominated by organic-carbon-rich sediments and functioned as net carbon sinks (on par with vegetated coastal habitats). These landscape-level differences reflect gradients in shellfish growth, survivorship and shell bioerosion. Notably, down-core carbon concentrations in 100- to 4000-year-old reefs mirrored experimental-reef data, suggesting our results are relevant over centennial to millennial scales, although we note that these natural reefs appeared to function as slight carbon sources (0.5 ± 0.3 MgC ha−1 yr−1). Globally, the historical mining of the top metre of shellfish reefs may have reintroduced more than 400 000 000 Mg of organic carbon into estuaries. Importantly, reef formation and destruction do not have reciprocal, counterbalancing impacts on atmospheric CO2 since excavated organic material may be remineralized while shell may experience continued preservation through reburial. Thus, protection of existing reefs could be considered as one component of climate mitigation programmes focused on the coastal zone.