Abstract. Surface melting over the Antarctic Peninsula (AP) plays a crucial role for the stability of ice shelves and dynamics of grounded ice, hence modulating the mass balance in a region of the world which is particularly sensitive to increasing surface temperatures. Understanding the processes that drive melting using surface energy and mass balance models is fundamental to improving estimates of current and future surface melting and associated sea level rise through ice-shelf collapse. This is even more important in view of both the paucity of in-situ measurements in Antarctica generally and the specific challenges presented by the circulation patterns over the Antarctic Peninsula. In this study, we evaluate the regional climate model Modèle Atmosphérique Régionale (MAR) over the Antarctic Peninsula (AP) at a 10 km spatial resolution between 1999 and 2009, a period which coincides with the availability of active microwave data from the QuikSCAT mission. This is the first time that this model, which has been validated extensively over Greenland, has been applied to the Antarctic Peninsula at a high resolution. We compare melt occurrence modeled by MAR with a combination of estimates from passive and active microwave data. Our primary regional focus is the northern East Antarctic Peninsula (East AP), where we evaluate MAR against wind and temperature data collected by three automatic weather stations (AWS). Our results indicate that satellites estimates show greater melt frequency, a larger melt extent, and a quicker expansion to peak melt extent than MAR in the center and east of the Larsen C ice shelf. The difference between the remote sensing and modeled estimates reduces in the north and west of the East AP. Melting in the East AP can be initiated by both sporadic westerly föhn flow over the AP and northerly winds advecting warm air from lower latitudes. To quantify MAR's ability to simulate different circulation patterns that affect melt, we take a unique approach to evaluate melt occurrence (using satellite data) and concurrent temperature biases associated with specific wind direction biases using AWS data over the Larsen Ice Shelf. Our results indicate that although MAR shows an overall warm bias, it also shows fewer warm, strong westerly winds than reported by AWS stations, which may lead to an underestimation of melt. The underestimation of föhn flow in the east of the Larsen C may potentially be resolved by removing the hydrostatic assumption in MAR or increasing spatial resolution. The underestimation of southwesterly flow in particular may be reduced by using higher-resolution topography.