Regional glacier mass balances can be measured by subtracting multi-temporal digital elevation models (DEMs). However, DEMs are often biased with altitude and it remains unclear whether the elevation differences observed on the ice-free terrain can be used to correct biases on ice-covered areas. IWe investigate such altitude-related biases using DEMs from three different sensors: SPOT5, SRTM C-band and SRTM X-band. The bias due to different original resolutions can be corrected using a relationship between curvature and elevation difference, calculated on ice-free terrain. The impact of C-band radar penetration into snow and ice can be evaluated for a specific region by comparing SRTM C-band and SRTM X-band DEMs. In our test area (Karakoram), the resolution-related bias has a minor influence on region-averaged elevation change. Conversely, not accounting for C-Band penetration would significantly biased the mean elevation change.