The future mass balance of the Antarctic ice sheet depends for an important part on the stability of the floating ice shelves that surround it, as these buttress the flow of grounded ice. Being situated relatively far north and near sea level, surface melt is a common but otherwise intermittent process on Antarctic ice shelves. Surface meltwater can form meltwater ponds, which can deepen existing crevasses that may eventually penetrate through the entire ice shelf. This process, called hydrofracturing, likely contributed to the recent disintegration of multiple ice shelves in the Antarctic Peninsula, most recently Larsen B ice shelf in 2002. A thorough understanding of surface melt processes is therefore key to improve our ability to predict future ice shelf stability and ice sheet mass loss. The snowmelt–albedo feedback plays a crucial role in Antarctic ice sheet surface melt: when snow melts, meltwater may refreeze in the snowpack, increasing the average grain size and lowering surface albedo. In turn, this enhances the absorption of solar radiation, further increasing surface melt rate. To investigate the importance of the snowmelt–albedo feedback for surface melt in Antarctica, we implemented an albedo parameterization in a surface energy balance model that calculates melt rates. In this parameterization, we can separate the impacts of dry and wet snow metamorphism on albedo evolution and melt rate. This allows us to quantify the snowmelt–albedo feedback, the results of which are presented here. Results for Neumayer Station on the Ekström ice shelf in Dronning Maud Land, East Antarctica, show that the snowmelt–albedo feedback results in a threefold increase in local melt rate compared to calculations in which the effect has been ignored. We also applied this method to weather station data from locations elsewhere in Antarctica. Finally, the same albedo parameterization was implemented in the regional climate model RACMO2. This provides the opportunity to extend this method to the entire Antarctic ice sheet, and to assess the temporal and spatial variability of the magnitude of the snowmelt–albedo feedback.