Abstract The Warm Arctic–Cold Eurasia (WACE) pattern is a key component of extratropical climate variability, shaping extreme weather across Eurasia. This study examines the historical variability and future projections of WACE using ERA5 reanalysis data, a 17-model CMIP6 multi-model ensemble, and a 50-member initial-condition ensemble from CanESM5 under the SSP5-8.5 scenario. Results indicate that both CMIP6 and CanESM5 ensembles effectively reproduce WACE’s spatial structure and variability. The pattern is primarily sustained by zonal and meridional heat transport and damped by the diabatic generation of available potential energy. Under global warming, WACE variability is projected to decline, with the most pronounced reductions in the Arctic. This decline is largely attributed to sea ice loss, increased ocean heat capacity, and enhanced upward turbulent heat flux, which collectively suppress Arctic temperature variability by weakening heat transport and intensifying diabatic heating. Model comparisons reveal that CanESM5 exhibits a stronger Arctic action center, larger zonal heat transport, and more pronounced diabatic heating damping than the CMIP6 ensemble, likely due to its higher climate sensitivity, leading to a sharper decline in Arctic temperature variance. Despite differences in the magnitude of projected changes and a typically larger inter-model spread compared to inter-member variability, the core processes linking changes in Arctic temperature variability to regional sea ice mass, surface heat fluxes, and horizontal heat transport are consistently represented across all models. These findings offer further insights into the evolving dynamics of extratropical climate variability in response to anthropogenic warming.