Ice wedges are a characteristic feature of northern permafrost landscapes and grow mainly by snowmelt that refreezes in thermal contraction cracks that open in winter. In high latitudes the stable-isotope composition of precipitation (δ18O and δD) is sensitive to air temperature. Hence, the integrated climate information of winter precipitation is transferred to individual ice veins and can be preserved over millennia, allowing ice wedges to be used to reconstruct past winter climate. Recent studies indicate a promising potential of ice-wedge-based paleoclimate reconstructions for more comprehensive reconstructions of Arctic past climate evolution. We briefly highlight the potential and review the current state of ice-wedge paleoclimatology. Existing knowledge gaps and challenges are outlined and priorities for future ice-wedge research are suggested. The major research topics are (1) frost cracking and infilling dynamics, (2) formation and preservation of the stable-isotope information, (3) ice-wedge dating, (4) age-model development and (5) interpretation of stable-isotope time series. Progress in each of these topics will help to exploit the paleoclimatic potential of ice wedges, particularly in view of their unique cold-season information, which is not adequately covered by other terrestrial climate archives.