Emerging responses of northern tree cover and composition under climate change have consequences for ecosystem functioning. Satellite-based global tree and land cover datasets have improved our understanding of northern tree cover dynamics. Yet, definitions of forests, design of retrieval algorithms, different spatial resolutions and satellite sensor quality can introduce uncertainties in these data. Here, our objective was to identify consistent patterns of recent change directions of tree cover and tree cover composition over northern lands using multiple datasets. We found large differences in area of tree cover changes in these datasets, ranging from 9% to 82% for increasing trends over North America and Europe. Consequently, we generated a synthesis map that captures consistent trends in tree cover in the four global datasets and found that the synthesis map showed higher agreement with visually interpreted tree cover changes from very high-resolution imagery . This result reflects improved consistency in areas of multi-dataset agreement, rather than improved absolute accuracy. The new synthesis map revealed a substantially larger area of increasing tree cover trends as compared to decreases. This pattern of net increases in tree cover was also relatively consistent among the major northern biomes, whereby large areas of increasing tree cover were particularly evident in boreal regions. Drier boreal and temperate biomes showed more areas with decreasing tree cover in comparison to wetter biomes. Our comparison of changes in the cover of evergreen and deciduous trees based on two land cover products uncovered large disagreements, making it nearly impossible at this point to attribute the identified tree cover changes in regards to forest type. These results may serve as a reference for model simulations of vegetation dynamics. Our findings also call for concerted efforts to produce more consistent tree cover and land cover datasets especially tree cover composition.