Experiments with dynamic-thermodynamic sea-ice models indicate a strong dependence of the net freezing rate, sea-ice transport and variability on dynamic model parameters. Although current dynamic-thermodynamic sea-ice models show relatively good agreement with observations, an optimization seems to be necessary: especially for the parameterizations of dynamic processes. Presently only a few coupled climate models use dynamic-thermodynamic sea-ice models. In order to promote, by means of coordinated numerical experiments, the development of an optimal sea-ice model for climate research, the Sea Ice Ocean Modelling Panel of the Arctic Climate System Study (ACSYS, a project of the World Climate Research Programme) has initiated the Sea ice model Intercomparison Project (SIMIP). The first results from this model hierarchy approach are presented.