Rising sea level and intensifying human disturbances are driving rapid changes of tidal wetlands worldwide. However, efforts to estimate their current and future status remain highly uncertain due to vigorous debate about how tidal wetlands—tidal marshes, mangroves and tidal flats—respond to drivers of change. We addressed this knowledge gap with a machine-learning analysis of more than 1 million satellite images to simultaneously map the extent, timing and type of change in the world’s tidal wetlands over the last two decades. We found that almost three-quarters of tidal wetland loss globally has been offset by the establishment of new tidal wetlands in areas where they formerly did not occur, suggesting that responses such as expansion, migration and restoration are widespread. Global analysis of a random sample of change pixels suggests that about 27% of all wetland changes were caused by direct human activities such as coastal development, with the remainder attributed to indirect drivers such as sea level rise and natural coastal processes. We hope our open-access data promotes the objective monitoring of coastal conservation and restoration efforts worldwide and contributes to a better understanding of the responses of tidal wetlands to accelerating global change.