This paper describes atmospheric general circulation model climate change experiments in which the Arctic sea-ice thickness is either fixed to 3 m or somewhat more realistically parameterized in order to take into account essentially the spatial variability of Arctic sea-ice thickness, which is, to a first approximation, a function of ice type (perennial or seasonal). It is shown that, both at present and at the end of the twenty-first century (under the SRES-A1B greenhouse gas scenario), the impact of a variable sea-ice thickness compared to a uniform value is essentially limited to the cold seasons and the lower troposphere. However, because first-year ice is scarce in the Central Arctic today, but not under SRES-A1B conditions at the end of the twenty-first century, and because the impact of a sea-ice thickness reduction can be masked by changes of the open water fraction, the spatial and temporal patterns of the effect of sea-ice thinning on the atmosphere differ between the two periods considered. As a consequence, not only the climate simulated at a given period, but also the simulated Arctic climate change over the twenty-first century is affected by the way sea-ice thickness is prescribed.