Saltmarshes provide crucial functions for flora, fauna, and humankind. Thus far, studies of their dynamics and response to environmental drivers are limited in space and time. Satellite data allow for looking at saltmarshes on a large scale and over a long time period. We developed an unsupervised decision tree classification method to classify satellite images into saltmarsh vegetation, mudflat and open water, integrating additional land cover information. By using consecutive stacks of three years, we considered trends while taking into account water level variations. We used Landsat 5 TM data but found that other satellite data can be used as well. Classification performance for different periods of the Western Scheldt was almost perfect for this site, with overall accuracies above 90% and Kappa coefficients of over 0.85. Sensitivity analysis characterizes the method as being robust. Generated time series for 125 sites across Europe show saltmarsh area changes between 1986 and 2010. The method also worked using a global approach for these sites. We reveal transitions between saltmarsh, mudflat and open water, both at the saltmarsh lower edge and interior, but our method cannot detect changes at the saltmarsh-upland boundary. Resulting trends in saltmarsh dynamics can be coupled to environmental drivers, such as sea level, tidal currents, waves, and sediment availability.