The Arctic Mediterranean (AM) is the collective name for the Arctic Ocean, the Nordic Seas, and their adjacent shelf seas. Water enters into this region through the Bering Strait (Pacific inflow) and through the passages across the Greenland–Scotland Ridge (Atlantic inflow) and is modified within the AM. The modified waters leave the AM in several flow branches which are grouped into two different categories: (1) overflow of dense water through the deep passages across the Greenland–Scotland Ridge, and (2) outflow of light water – here termed surface outflow – on both sides of Greenland. These exchanges transport heat and salt into and out of the AM and are important for conditions in the AM. They are also part of the global ocean circulation and climate system. Attempts to quantify the transports by various methods have been made for many years, but only recently the observational coverage has become sufficiently complete to allow an integrated assessment of the AM exchanges based solely on observations. In this study, we focus on the transport of water and have collected data on volume transport for as many AM-exchange branches as possible between 1993 and 2015. The total AM import (oceanic inflows plus freshwater) is found to be 9.1 Sv (sverdrup, 1 Sv =106 m3 s−1) with an estimated uncertainty of 0.7 Sv and has the amplitude of the seasonal variation close to 1 Sv and maximum import in October. Roughly one-third of the imported water leaves the AM as surface outflow with the remaining two-thirds leaving as overflow. The overflow water is mainly produced from modified Atlantic inflow and around 70 % of the total Atlantic inflow is converted into overflow, indicating a strong coupling between these two exchanges. The surface outflow is fed from the Pacific inflow and freshwater (runoff and precipitation), but is still approximately two-thirds of modified Atlantic water. For the inflow branches and the two main overflow branches (Denmark Strait and Faroe Bank Channel), systematic monitoring of volume transport has been established since the mid-1990s, and this enables us to estimate trends for the AM exchanges as a whole. At the 95 % confidence level, only the inflow of Pacific water through the Bering Strait showed a statistically significant trend, which was positive. Both the total AM inflow and the combined transport of the two main overflow branches also showed trends consistent with strengthening, but they were not statistically significant. They do suggest, however, that any significant weakening of these flows during the last two decades is unlikely and the overall message is that the AM exchanges remained remarkably stable in the period from the mid-1990s to the mid-2010s. The overflows are the densest source water for the deep limb of the North Atlantic part of the meridional overturning circulation (AMOC), and this conclusion argues that the reported weakening of the AMOC was not due to overflow weakening or reduced overturning in the AM. Although the combined data set has made it possible to establish a consistent budget for the AM exchanges, the observational coverage for some of the branches is limited, which introduces considerable uncertainty. This lack of coverage is especially extreme for the surface outflow through the Denmark Strait, the overflow across the Iceland–Faroe Ridge, and the inflow over the Scottish shelf. We recommend that more effort is put into observing these flows as well as maintaining the monitoring systems established for the other exchange branches.