One of the greatest challenges in atmospheric models is the correct simulation of supercooled liquid water in clouds, and more particularly of the relative amounts of supercooled liquid and ice in mixed-phase clouds. In the polar regions these clouds are ubiquitous. Our understanding of the underlying processes leading to their formation is incomplete and the observations keep challenging our ability to model them. This is problematic given the strong impact of mixed-phase clouds on the surface radiative budget, and the known polar amplification through which the Antarctic and the Arctic are undergoing stronger warming than anywhere else on Earth. In the Southern Ocean down to the Antarctic coasts, climate models display the largest biases in the net cloud radiative effect, and this traces back to their inability to correctly account for the supercooled liquid water. One reason for this is our poor understanding of cloud-aerosol interactions in such a remote region where natural marine aerosols play a key role in controlling cloud formation. Contrary to the Arctic where anthropogenic pollution and continental aerosols complicate aerosol-cloud interactions, the Antarctic can be seen as a natural laboratory with pre-industrial conditions where key polar-specific processes can be studied, which involve interactions with sea ice and marine aerosols - whether mineral or biogenic. We will present observations of mixed-phase clouds as well as cloud modelling results in the Antarctic. On one hand we will explain how aircraft in-situ observations allow us to point at critical microphysical processes taking place in clouds and the links with the surrounding environment and – on the other hand - why satellites and more particularly radar and lidar observations are necessary to constraint the cloud microphysical properties Antarctic-wide, backed up by ground-based observations. Modelling results will be presented to highlight the standing issues in capturing the variability of the occurrences and of the liquid and ice contents of polar clouds, and the remaining mysteries about the actual sources of the cloud condensation nuclei and the ice nucleating particles responsible for their formation. Overall this has important consequences regarding the current efforts to correctly model the surface energy balance and the surface ice mass balance in the polar regions.