AB: Though a number of studies have reported CO2 fluxes from the Arctic, few of these include measurements from winter time and it is ften assumed that emission rates during winter time are either constant or negligible. These assumptions are often made because no data are available or consist of relatively few measurements which appear to give small and constant emission rates. Further, most studies of the processes behind winter time emission of CO2 conclude that the flux during this time of year can be linked to the respiratory release of CO2 from soil micro organisms, which is temperature dependent and given the low temperatures of the Arctic winter it is reasonable to assume that CO2 production is low. Here we present measurements from a new design of an open top chamber which has show a good potential for continuous measurements of CO2 and other gases even under extreme coldness during the winter at a high arctic location in Svalbard (78°N). Measurements were conducted in the field during the winter season of 2004-2005 and show reliable and continuous measurements of CO2 fluxes down to a level of 0.01 ìmol m-2 s-1 and good correspondence with other types of soil chambers. Our results indicate that a substantial part of the annual CO2 emission from the ecosystem occur during the freeze in period, where more CO2 is emitted from the soil over a few weeks than the accumulated flux for the rest of the winter. During the coldest part of the