Recent high-quality hydrographic measurements in the Arctic Ocean have revealed a warming of up to 1 K of the Atlantic Layer when compared with Russian climatologies of the 1940s to 1970s. About half of this warming can be attributed to the different methods by which the two data sets were obtained: the climatologies are based on discrete bottle data in the vertical and necessarily involve smoothing in space and time, whereas the modern, quasi-synoptic data are from continuously recording conductivity-temperature-depth sondes and give much better resolution in the vertical. Modern surveys also focused more on boundary current regimes where narrow warm bands of Atlantic Water are present. The remainder of the warming detected can be explained with the following physical arguments: (1) an increased inflow of Atlantic Water in the early 1990s and (2) a higher temperature of this inflowing water. Temperature time series in the Barents Sea since the beginning of this century suggest that the warming of the early 1990s is not a long-term climate signal but, rather, is related to the inherent natural variability of the system with timescales of decades.