Recent experiments have shown that low frequency ultrasound (LFUS) induces leakage from lipid vesicles. However, the mechanism by which LFUS disrupts the lipid bilayer structure is not clear. In this paper we develop a theoretical model to test the possibility that gas molecule partitioning from the aqueous media into the lipid bilayer core can lead to the nucleation of microscale gas bubbles. If those can, indeed, form, then their presence in the lipid bilayer and interactions with an ultrasound field can cause bilayer disruption and leakage. The model derived here for the nucleation of stable bubbles accounts for the ‘surface tension’ that the lipid bilayer exerts on the bubble, a result of the associated disruption of the lipid packing. The model predicts that the probability of bubble nucleation is highly sensitive to the bilayer thickness, and largely insensitive to the bilayer phase. The probability of stable bubble formation is shown to correlate with experimentally measured sensitivity of lipid bilayers to LFUS, suggesting that membrane disruption may be due to embedded bubbles that nucleated in the bilayer.