European winters still perceived as cold, even under strong warming

During several periods of winter 2025/2026, some European regions experienced relatively harsh winter weather, with low minimum air temperatures. Society has perceived these conditions as extreme. However, objectively, the temperature minima were far from extremes recorded during the 20th century. With ongoing warming, even warmer cold extremes might result in considerable impacts across sectors and natural ecosystems. Here, we investigate how changes in the internal variability of winter temperature might alter the shape of its statistical distribution under stronger radiative forcing, with implications for extremes. We explore two different sets of climate model runs: LongRunMIP simulations, comparing near-equilibrium conditions under preindustrial and abrupt 4xCO2 forcings, and transient large ensemble simulations comparing the historical and scenario periods (the end of the 21st century under RCP8.5/SSP5-8.5 socio-economic pathways). A change in the shape of the temperature distribution can then point to a fundamental change in climate-governing processes. In agreement with previous studies, we reveal that a decrease in variance will accompany increasing winter mean temperatures, as day-to-day temperature variations are induced by the occurrence of synoptic-scale weather systems, and in warmer climates, this is expected to decline. Our study provides new insights, showing that the variance shrinking is spatially heterogeneous. We further focus on the skewness of the temperature distribution and investigate changes in the lengths of the cold and hot tails, which are related to changes in variance. In central, western, and northern Europe, skewness is decreasing, and the cold tail is shrinking more slowly than the hot tail, implying enduring cold extremes, even in climatic states much warmer than those we are familiar with. On the other hand, the results for the Mediterranean are almost opposite. We discuss the results in the context of recently observed temperature evolution.