This paper explores the origins and evolution of ice-rich interior mounds found within craters of the north polar region of Mars. We present a systematic study of impact craters above 65°N, and identify 18 craters that have interior mounds. At least 11 of these mounds are composed of water ice and geometric similarities suggest that dune-covered mounds may also have a water ice core. The mounds are found in the deeper craters in the north polar area and we suggest that these form a specific microclimate favorable for mound initiation and growth. It is likely that at least seven of the mounds have evolved as individual outliers, rather than conterminous with the main polar cap. Our observations suggest that the mounds are built up by atmospheric deposition, similar to that of the north polar layered deposits. Using a combination of remote sensing techniques enabling topographic, spectral, radar and image data analyses, we have documented the morphology, composition and stratigraphy of selected mounds. We advance and test four hypotheses for formation of these mounds: artesian outpouring from a deep aquifer, hydrothermal activation of ground ice, remnants of a more extensive polar cap, and atmospheric deposition on ice caps in meteorologically isolated locations. We propose that during periods when the perihelion was located in northern summer (most recently 10–25 ka before present) the microclimate in these craters retarded the sublimation of CO2 and water ice in northern spring, thus creating a cold trap for volatiles released as the seasonal cap retreated. This created a thick enough deposit of water ice to withstand sublimation over the summer and initiate a positive feedback leading to mound-building. Mounds without complete dune-cover may be in dynamic equilibrium with the ambient climate and show evidence of both present-day and past periods of erosion and aggradation. We conclude that the water ice mounds formed in deep impact craters in Mars’ north polar region may contain sensitive records of past polar climate that may enhance our understanding of the CO2–H2O system in the polar regions.