This study considers the possible use of different kinds of forcing datasets in Baltic Sea ocean climate modelling on centennial time scales, in particular for the past half millennium. We demonstrate that high-quality station data of the past century and gridded multi-proxy reconstructions for the past 500 years can be used with great success but with various levels of detail. We also demonstrate that output data from the state-of-the-art global climate model EcHo-G are not suitable for modelling the Baltic Sea ocean climate. Two climate properties were studied: the annual maximum ice extent (MIB) and the vertically and horizontally integrated annual water temperature. Centennial time scale results indicate that the seventeenth and nineteenth centuries were the coldest centuries, while the 1690s were the coldest decade and 1695 the coldest year in the last 500 years. The results also indicate that the twentieth century was the warmest century with the least MIB of the last 500 years. On a decadal time scale, the 1990s, 1930s and 1730s were the warmest decades and comparable in terms of both water temperature and MIB. The year 1989 had the minimum observed MIB of only 52,000 km², implying that the Baltic Sea has been partly ice covered in all winters of the past half millennium. Even though different climate forcing mechanisms may operate on the climate system today compared to over the last half millennium, this study cannot clearly state that the region is experiencing climate change outside the natural limits of the past 500 years.