The assessment of large-scale land cover pattern is an important input for monitoring and modeling of ecological and environmental processes. Considerable efforts have recently resulted in the development of global continuous fields for different land cover types at large spatial scales based on NOAA-AVHRR and TERRA-MODIS data. In this study, a new methodology is described for deriving continuous fields of tree cover (coniferous/deciduous) and vegetation not covered by trees as well as non-vegetated land cover for complex topography at the regional scale of Switzerland. The methodology is based upon generalized linear models (GLM). MODIS (MOD09A1) data at a spatial resolution of 500 m was used to calibrate the GLM models. For purpose of validation we compared the resulting continuous fields of tree cover and an available global data set, namely the TERRA-MODIS Vegetation Continuous Fields product (MOD44B) against an independent reference dataset. In comparison with the available global tree continuous field data set (MOD44B), our regional GLM-based tree model obtained significantly better results for all calculated accuracy measures. Tests of the resulting maps showed that non-vegetated, bare ground and vegetation covered by trees were predicted accurately, with weighted Kappa values (K_w) reaching 0.87 and 0.85 respectively. Due to the heterogeneous character of "non-tree vegetation" this latter landcover type was predicted with lower precision (K_w = 0.79). We conclude that generalized linear models are appropriate for deriving continuous fields of different land cover types for complex topography at a regional scale. Regional calibration of land cover continuous fields offers significantly improved predictions compared to globally calibrated models and may serve as a valuable tool for regional monitoring of land cover pattern and its temporal change.