Multilayer clouds (MLCs) are vertically stacked cloud systems that occur frequently in the Arctic and strongly influence the region's surface energy balance. Yet the air mass conditions driving their occurrence and vertical structure remain poorly understood. Here, we investigate 19 years of wintertime MLC observations at Ny-Ålesund, Svalbard, an Atlantic-sector Arctic gateway subject to frequent warm air intrusions (WAIs) and cold air outbreaks (CAOs). We use radiosonde profiles to identify potential cloud layers and verify them using collocated cloud radar. Air mass regimes are classified from equivalent potential temperature profiles using rotated principal component analysis. The radiosonde profiles show that thermodynamic conditions favorable for MLCs are common, with 54.4% of winter profiles containing multiple potential cloud layers. Radar verification reduces this to 27.3%, as some potential layers lack hydrometeors, particularly in the upper troposphere. Across air mass regimes, WAIs are the cloudiest regime and favor vertically complex cloud structures, with 43.8% of radar-verified WAI profiles containing multiple layers compared with 21% during CAOs. Idealized seeder-feeder calculations indicate that falling small ice particles are more likely to reach the lower cloud during CAOs, whereas for larger particles, vertical layer separation is the main constraint regardless of synoptic conditions. Radiatively, MLCs modulate surface longwave cooling most effectively during CAOs, while during WAIs single- and multilayer clouds produce similar effects. Thus, the air mass regime, rather than the number of cloud layers, modulates how MLCs affect the Arctic winter surface energy budget, providing an observational basis for evaluating MLCs in models.