Abstract. Characterizing hydrological processes within tropical wetlands is challenging due to their remoteness, complexity and heterogeneity. In particular, estimates of evaporative water loss are inherently uncertain. In view of the large influence on the local and regional climate, the quantification of evaporation is essential for the determination of the water balance of permanent and intermittent water bodies. Data for tropical wetlands are scarce where their remoteness impedes direct evaporation measurements. Seasonal inundation dynamics affect evaporation processes in tropical wetlands, which can be analysed in two stages: the first stage during the wet season and the second stage during the dry season. As yet no adequate method exists for determining second-stage evaporation in a data-scarce environment that additionally allows for a transfer of simulated actual evaporation (AET) to other locations. Our study aimed at developing a process-based model to simulate first-and second-stage evaporation in tropical wetlands. We selected a set of empirical potential evaporation (PET) models of varying complexity, each based on different assumptions and available data sets, and evaluated the models with pan evaporation observations in the Pantanal of South America, one of the largest tropical wetlands in the world. We used high-resolution measurements of surface and groundwater levels at different locations to determine the water available for evaporation. AET was derived by constraining simulated PET based on available water. The model of best fit was applied to different types of water bodies with varying hydroperiods to capture first-and second-stage evaporation across a range of wetland types. With our new model we could quantify evaporative water loss in the dry and the wet season for different locations in the Pantanal. This new spatially explicit approach represents an improvement in our understanding of the role of evaporation in the water balance of the Pantanal. We recommend the application of this model in other remote tropical wetlands, since only a minimum of input data is necessary.