Within SWECLIM a 3D fully coupled ice-ocean model has been developed based on the massively parallel OCCAM code from Southampton. Compared to the global OCCAM the model has to be adopted to Baltic Sea conditions with implementations of high-frequent atmospheric forcing fields in connection with adequate bulk formulae for wind stress, heat uxes and freshwater uxes, solar radiation, river runoff, active open boundary conditions, a second-order moment turbulence closure scheme and a dynamic-thermodynamic sea ice model. Thereby, state-of-the-art sub-models and parameterizations have been used. RCO is the first 3D coupled ice-ocean model for the Baltic Sea with the above mentioned specifications suitable for use on mpp computers like CRAY-T3E's. Thus, a milestone for 3D ocean model development has been set. No other model is as fast as RCO. The performance has been improved significantly using advanced algorithms to optimize processor maps. This guarantees work load balance between the different processors. From now on it is possible to perform longterm simulations (10 years) within SWECLIM using a sufficiently resolved 3D Baltic Sea model. The open boundary conditions have been tested. They allow waves to radiate out of the model domain and signals prescribed at the border to in uence the model interior. No significant trends (like emptying or filling) have been observed which might prevent longer integrations of the system. An option has been included in RCO for active open boundary conditions also for temperature and salinity. For the first time the turbulence closure model has been tested within a 3D model in all Baltic sub-basins. The new flux boundary conditions for turbulent kinetic energy parameterizing breaking surface waves perform well. First results for the hindcast period 1992/93 are presented. Therefor, realistic atmospheric, runoff and boundary data have been used. The model is initialized using observed profile temperature and salinity data. A spin-up period of 3 months starting in May is sufficient to smooth out artificial gradients from the initialization procedure and to turn in basin wide volume changes correctly. The model results have been compared to sea level, sea surface temperature, temperature/salinity profile and ice thickness/compactness data with good agreement. Basin wide volume changes as well as daily sea level oscillations are simulated surprisingly good. Sea surface temperatures follow the observed seasonal cycle. Up- and downwelling events in RCO occur as observed with the right frequency and area extent but the sst's tend to be colder in upwelling and warmer in downwelling regions compared to observations. Mixed layer depths, which are important for the ocean heat content, agree well with previous model studies which are validated against observations intensively (Meier, 1996). The water exchange between Baltic and North Sea crucial for multi-year integrations is modelled realistically. Especially the salt water inflow in January 1993 can be reproduced. The bottom water in Bornholm Basin is replaced by new water originating from the North Sea but maximum observed bottom salinities at Bornholm Deep are underestimated by 1-2 PSU. Freezing, breakup date and maximum ice extent are in good correspondence with observations. Improved parameterizations result in modelled ice thicknesses as observed whereas other authors report too large ice thicknesses and delayed ice melting (e.g., Haapala and Lepparanta,1996). Multi-year simulations including mild, normal and severe winters will be necessary to elucidate this problem further. A comparison between an experiment with full dynamic-thermodynamics and one without dynamic effects reveals the importance of ice advection under wind influence. A process study from the beginning of February 1993 showed that under strong wind conditions a hole in the ice coverage can open with the size of half of the Bothnian Bay. At the end of January 1993 the Bothnian Bay, the coastal area of the Bothnian Sea and the eastern parts of the Gulf of Finland are ice covered. A couple of days later westerly winds led to wide open areas in the western Bothnian Bay while ice piled up at the eastern coasts to a correct amount. This phenomenon can be modelled only with ice dynamics included. The aim of SWECLIM is to increase our knowledge of the effects of climate change in Sweden and the other Nordic countries. Therefor, it is necessary to understand the present climate. For the Baltic Sea even the knowledge about the present mean state and its transients is rather poor. Only a small number of long-time observations like sea level records (for example from Stockholm, see Ekman (1988)), maximum annual ice extent (e.g., Palosuo, 1953; Seina and Palosuo, 1993) or temperature and salinityprofiles from monitoring stations in some of the sub-basins (e.g., Matthaus and Frank,1992) are available. These informations are not enough to understand the driving mechanisms of mean horizontal and vertical transports of energy, momentum and matter. 3D Baltic Sea models like RCO will close this knowledge gap in future. Thereby, it will be possible to close the water and energy cycle of the Baltic catchment area, a final goal of BALTEX. By applying atmospheric forcing data from scenario simulations in one- or two-way coupled mode it will be possible to make predictions of climate change for the Baltic Sea. Impact studies of the future marine environment will be available using detailed highly resolved information from RCO. This report presents a powerful tool for solving these and other tasks.