This presentation aims at the parameterization of the atmospheric drag coefficient over sea ice and on linking it to retrievals of surface roughness parameters. The sea-ice roughness is quantified by using sea-ice surface elevation profiles obtained from an airborne laser altimeter and by using optical and SAR images. Data acquired during three different aircraft campaigns are used. All flights took place in the vicinity of Svalbard (over the Barents Sea, the Storfjord and Fram Strait) in spring 1998, 2005 and 2007, respectively. Different types of ice were detected (ridged and smooth first-year ice, broken ice and nilas). Ice regimes did not only vary between different test sites but also along the flown profiles. The atmospheric drag coefficients are derived from the wind components measured at high frequency. The values were compared with values derived from parameterizations using two different approaches. The first approach uses the concept of drag partitioning. This concept distinguishes between the influence of small-scale roughness (skin drag) and of larger distinct obstacles (form drag) on the total surface roughness. The drag coefficient can be expressed as a function of mean ridge heights and spacings averaged over flight legs of 10 km. We found a good agreement between parameterized and observed drag coefficients. To overcome some ambiguity resulting from the used ridge definition another approach was tested which uses statistical properties of the sea-ice surface topography (like the variance) to calculate drag coefficients. First results indicate also a good agreement between measured and parameterized drag coefficients.