Understanding the evolution of Arctic sea-ice is crucial due to its climatic and socio-economic impacts. Usual descriptors (e.g., sea-ice extent, sea-ice age, and ice-free duration) quantify changes but do not account for the full seasonal cycle. Here, using satellite observations of sea-ice concentration over 1979–2023, we perform a k-means clustering of the Arctic sea-ice seasonal cycle, initializing with equal quantile separation and using Mahalanobis distance. We identify four optimal seasonal cycle clusters: open-ocean (no ice year-round), permanent sea-ice (full coverage with a minimum of 70 % sea-ice concentration), and two clusters showing ice-free conditions, namely partial and full winter freezing. The latter has larger sea-ice concentration in winter, more abrupt melting and freezing periods, and a shorter ice-free season than the former. The probability of belonging to the open-ocean cluster increased by 1.6 % per decade mostly due to cluster spatial expansion on the Eurasian side. The permanent sea-ice decreased by 1.5 % per decade with a likelihood reduction in the Canadian side. The partial and full winter freezing clusters do not exhibit any trend but spatial shifts occur. We further diagnose cluster transitions and subsequently infer regions of stabilization and destabilization. The East Siberian and Laptev seas are destabilizing (losing their typical permanent sea-ice seasonal cycle) while the Kara and Chukchi seas have stabilized (experiencing a new typical seasonal cycle, corresponding to the partial winter-freezing cluster). This work provides a new way to describe Arctic regional changes using a statistical framework based on physical behaviours of sea-ice.