Sea ice is an active source or a sink for carbon dioxide (CO 2 ), although to what extent is notclear. Here, we analyze CO 2 dynamics within sea ice using a one-dimensional halothermodynamic sea ice model including gas physics and carbon biogeochemistry. The ice-ocean fluxes, and vertical transport, of total dissolved inorganic carbon (DIC) and total alkalinity (TA) are represented using fluid transport equations. Carbonate chemistry, the consumption, and release of CO 2 by primary production and respiration, the precipitation and dissolution of ikaite (CaCO 3 Á6H 2 O) and ice-air CO 2 fluxes, are also included. The model is evaluated using observations from a 6 month field study at Point Barrow, Alaska, and an ice-tank experiment. At Barrow, results show that the DIC budget is mainly driven by physical processes, wheras brine-air CO 2 fluxes, ikaite formation, and net primary production, are secondary factors. In terms of ice-atmosphere CO 2 exchanges, sea ice is a net CO 2 source and sink in winter and summer, respectively. The formulation of the ice-atmosphere CO 2 flux impacts the simulated near-surface CO 2 partial pressure (pCO 2 ), but not the DIC budget. Because the simulated ice-atmosphere CO 2 fluxes are limited by DIC stocks, and therefore <2 mmol m 22 d 21 , we argue that the observed much larger CO 2 fluxes from eddy covariance retrievals cannot be explained by a sea ice direct source and must involve other processes or other sources of CO 2 . Finally, the simulations suggest that near-surface TA/DIC ratios of $2, sometimes used as an indicator of calcification, would rather suggest outgassing.