Future changes in interannual variability (IAV) of Arctic climate indicators such as sea ice and precipitation are still fairly uncertain. Alongside global warming-induced changes in means, a thorough understanding of IAV is needed to more accurately predict sea ice variability, distinguish trends and natural variability, as well as to reduce uncertainty around the likelihood of extreme events. In this study we rank and select CMIP6 models based on their ability to replicate observations, and quantify simulated IAV trends (1981–2100) of Arctic surface air temperature, evaporation, precipitation, and sea ice concentration under continued global warming. We argue that calculating IAV on grid points before area-averaging allows for a more realistic picture of Arctic-wide changes. Large model ensembles suggest that on shorter time scales (30 years), IAV of all variables is strongly dominated by natural variability (e.g. 93% for sea ice area in March). Long-term trends of IAV are more robust, and reveal strong seasonal and regional differences in their magnitude or even sign. For example, IAV of surface temperature increases in the Central Arctic, but decreases in lower latitudes. Arctic precipitation variability increases more in summer than in winter; especially over land, where in the future it will dominantly fall as rain. Our results emphasize the need to address such seasonal and regional differences when portraying future trends of Arctic climate variability.