Abstract The horizontal water vapour transport (IVT) is one of the key variables connected to precipitation extremes and floods, especially in the mid-latitudinal countries, and represents a key link between water sources and sink regions. Because of its significant impacts, great efforts have been made to examine IVT and its projected changes in response to changes to the climate system, leveraging outputs from global climate models (GCMs). However, to gain more confidence in the projections, it is important to evaluate how well the GCMs can reproduce the historical past. Here, we assess how well 10 GCMs archived in the Climate Model Intercomparison Project Phase 5 can reproduce the spatial and temporal patterns of the mean and 85th percentile of the IVT distribution. Analyses are performed at the global scale using outputs at the 6-hourly resolution, and four different reanalysis products as reference. We find that the results from the GCMs are in good agreement with the reanalyses in terms of the IVT climatology; however, the GCMs fail to capture the trends in IVT, with almost all the models showing a negative correlation with the reanalysis data for both the mean and the 85th percentile. We also extend our analyses to the daily time scale, reaching the same conclusions drawn for the 6-hourly resolution. These results further highlight the need for a careful evaluation of the GCM outputs in reproducing the observed IVT distribution.