Freeze-induced stress causing aggregation of proteins has typically been primarily attributed to the ice-water interface. However, we hypothesize that the underlying observed and perceived detrimental effect of ice is, to some extent, attributed to air bubbles expelled from ice crystal lattices or to nanobubbles existing prior to freezing. The reduction of dissolved air was achieved via a deaeration process by placing samples in a reduced pressure chamber, while the reduction of nanobubbles was achieved by filtering samples via a syringe filter. The results showed that the reduction of both dissolved air molecules and stable colloidal nanobubbles in a bovine IgG solution prior to freezing led to a significant decrease in aggregation after thawing compared to untreated samples (∼6,000 vs. ∼ 40,000 particles/mL at a freezing rate of 100 K/s, respectively). The deaeration-filtration treatment works additively with cryoprotectants such as trehalose, further reducing the freeze-induced aggregation of IgG. The results also demonstrated that air–water interfacial aggregation of IgG in bulk liquid samples is a time-dependent process. The number of IgG subvisible particles increased with time and temperature, suggesting that random collisions of denatured molecules promoted the formation of aggregates with spherical morphology. In contrast, the IgG subvisible count after freeze-thawing had already reached its nominal value, suggesting a time-independent process where denatured protein molecules were compressed between ice crystals into filament-like aggregates. In summary, the findings shift the current paradigm from ice crystals being the main destabilizing factor during freezing to air bubbles, although the two are intertwined. From a translational aspect, this study underscores the value of deaeration-filtration as an essential supplemental process that can be applied in addition to formulation approaches such as the use of cryoprotectants to further reduce freezing stress on proteins and increase their stability.