A Baltic dataset covering the period 1990–2020 is reconstructed using a high-resolution model and data assimilation. Satellite sea surface temperature observations and temperature and salinity (T/S) profiles are used to reduce the model biases by a local singular evolutive interpolated Kalman filter. The dataset is evaluated with assimilated T/S profiles, reprocessed grid observations, and digitized ice charts. The results show that the sea surface temperature, the mixed layer depth, the vertical distribution of T/S, and sea ice are all reasonably reproduced. T/S biases are greater in the coastal waters than in the open waters. T/S trends on the surface and the bottom of the Baltic Sea are analyzed for the first time from a reanalysis perspective. Generally, the Baltic Sea shows a clear warming trend in recent decades, with a slight desalination trend in the northern Baltic Sea and a salty trend in the southern Baltic Sea. This study also demonstrates the Baltic T/S trends are larger in the south than in the north. Temperatures appear the largest trend in the Gulf of Finland surface and in the eastern Gotland Basin bottom, whereas salinities show the largest trend in the Danish Straits surface and in the northern Gotland Basin bottom, respectively. Furthermore, T/S trends in the Baltic Sea show different temporal features. Baltic surface and bottom temperature trends over the past 30 years are at their largest during 2010–2020. Surface salinity discrepancies between recent decades are clear in winter and spring beginning.