A telescope polarization model for the SST [Swedish 1-m Solar Telescope] is developed and the parameters of this model are fitted to polarization measurements made with a 1-meter linear polarizer in front of the entrance window. In this model, the 1-meter lens is characterized by a five-parameter Muller matrix, corresponding to a retarder with arbritary variations of the retardance and fast-axis orientation across the aperture. The resulting model is verified by measuring the telescope polarization for unpolarized input light and comparing to predictions from the polarization model. The accuracy of the prediction is within approximately 0.4% for all normalized polarization components (Q/I, U/I and V/I). The polarimeter used is based on two nematic liquid crystals and one linear polarizer, and will be used for both imaging polarimetry and spectropolarimetry. The most critical calibration is measuring the modulation matrix. This is done by inserting one linear polarizer and one rotating quarter-wave plate in the optical path before the polarimeter, and measuring the modulated intensity. The calibration of the quarter-wave plate is optimized by measuring the linear polarizer only with the polarimeter, and then minimizing the error in degree of polarization plus the residual error for the inversion of the modulation matrix by iteration of the two unknown parameters (retardance and angle offset). We find that small non-linearities in the CCD response is the major obstacle in calibrating the polarimeter. The first full Stokes imaging polarimetry observations at the SST are shown. Comparing images before and after telescope compensation verify the telescope polarization model.