Earth’s violent accretion likely generated multiple magma oceans. In particular, the Moon-forming giant impact is often thought to have produced a whole mantle magma ocean, which would have homogenized any pre-existing chemical heterogeneity within the mantle. The ratio of primordial 3He to primordial 22Ne in the mantle preserves a record of magma oceans on the early Earth. Importantly, the 3He/22Ne ratio of the Earth’s shallow depleted mantle is significantly higher than the deep mantle. To explain this observation, I propose that at least two giant impact-induced atmospheric blow-off and magma ocean degassing episodes are required and that the last giant impact did not generate a whole mantle magma ocean. New Xe isotopic data indicate that the catastrophic mantle outgassing and atmospheric blow-off events inferred from3He/22Ne ratios were accomplished between ~30 to 55 Myrs after the start of the Solar System. Therefore, outgassing associated with giant impacts, including the Moon-forming impact, must have occurred within this time window. Previous calculations of impact-induced atmospheric erosion have, however, found that it is difficult to completely remove the atmosphere from a body as large as Earth by a giant impact. The need for atmospheric loss inferred from the noble gas data could be reconciled with the dynamics of giant impacts by considering the new high-spin Moon formation hypothesis. I will further show that the current inventory of primordial noble gases in the atmosphere must largely be derived from late accreting planetesimals, a conclusion that has implications for the composition of the early atmosphere.