This paper re-evaluates the data for inner Solar System volatiles with particular reference to the Earth. The mass balance afforded by 40Ar/36Ar shows that the mantle as sampled by volcanism contains at most a small proportion (1-3%) of Earth's primordial argon regardless of the exact K/U. This mass balance is derived from MORB, OIB and well gases. Assuming it represents the total mantle therefore, it can be combined with estimated MORB- and OIB-source budgets to derive a ratio of (seismic) lower to upper mantle primordial noble gas concentrations of 6.9 ± 5.6. The upper and lower mantle concentrations can be made to balance if there have been major (∼40%) losses of highly incompatible elements by impact erosion and the K/U of the MORB source is high (19,000) as recently proposed. Both impact erosion and lower K/U serve to reduce the 4.0 Ga apparent K-Ar age of the mantle, which would be more consistent with significant levels of K and noble gas recycling over geological time. Using noble gases, two extreme models are derived for the H, C and N budgets of Earth's mantle: a layered mantle model, and an impact erosion (uniform) mantle with a composition like that of the MORB source. The impact erosion model better replicates the budgets derived from direct measurement of H, C and N in basaltic glasses but how representative these are of the lower mantle is unknown. These models are independent of the ultimate origins of the noble gases, which are evaluated using non-radiogenic ratios. The 20Ne/36Ar, 20Ne/22Ne and 36Ar/38Ar of Earth, Venus and Mars are consistent with derivation from chondritic materials with admixed Solar components. The Solar proportions of Ne in Earth's atmosphere (∼20%) and mantle (∼75%) are used to derive a likely 3He budget of 4.0 × 1038 atoms for the primordial atmosphere. The heavy noble gases are inconsistent with these simple mixtures and present clear evidence of a major component derived from amorphous cometary ices fractionated from Solar and CI-like compositions that could contribute about 20-50% of the Kr in the atmosphere and potentially more in the mantle. The heavy noble gases in the mantle are not just elementally fractionated but also include Xe that is isotopically heavy, like the atmospheric Xe in Earth and Mars. Therefore, the mantle probably includes protoplanetary and early atmospheric noble gases with cometary and EUV-fractionated components incorporated during accretion and/or by subduction. Earth's Solar normalised primordial abundances of 1H, 3He (determined from the 36Ar mass balance), 12C, 14N, 20Ne, 36Ar, 84Kr and 130Xe, all ignoring the core, correlate with those in chondrites. Primordial 3He, 20Ne, 36Ar and 84Kr proportions are especially close to chondritic but are two orders of magnitude lower in abundance than those of Venus. This may reflect bulk loss of the atmosphere during the Moon-forming Giant Impact. Assuming CI chondrites are Earth's main starting materials for volatiles, 1H is as depleted as 130Xe, and 12C and 14N are the most depleted stable elements in the bulk silicate Earth. The most highly volatile elements 3He, 20Ne, 36Ar and 84Kr are two orders of magnitude more abundant, and are less depleted even than the most highly siderophile elements (PGEs, Re, Au, Te, Se and S), commonly used to define the mass of a late veneer. The inferred amorphous ice cometary noble gas contributions cannot explain the budgets of 1H, 12C, 14N; these can only be derived from chondrites otherwise noble gas budgets would be far higher. A veneer of chondritic material with a minor amount (10-30 ppm) of admixed model cometary composition would explain the noble gas elemental proportions and their overall budget relative to C. However, Earth's H/C and C/N neglecting unknown core contributions are strongly non-chondritic and inconsistent with any combination of chondritic or cometary materials. If a late chondritic veneer contributed most of Earth's nitrogen more than 70% of the hydrogen, presumably in the form of water, would need to predate it. Therefore, Earth probably acquired volatile elements from chondritic material admixed with Solar and cometary contributions during the main stages of accretion, but this was accompanied or followed by greater but variable depletion in 1H, 12C, 14N and 130Xe possibly supplemented by the addition of a late veneer. Venus and Mars display a broadly similar pattern of C and N depletion relative to noble gases when chondrite normalised, based on the minimum budgets deduced from their atmospheres. The strong depletion of 1H, 12C, 14N and 130Xe relative to other noble gases in terrestrial planets, and possibly Xe isotopic fractionation as well, could be explained by the early removal of these elements from the inner circumstellar disk, from the planets, or from silicate reservoirs themselves. Some of the lost 1H, 12C, 14N and possibly 130Xe could be in the metallic cores of terrestrial planets. However, carbon, nitrogen and xenon also all form low temperature species with ionization potentials less than that of hydrogen. The depletion of these four elements as well as the strong Xe isotopic fractionation may therefore also relate to loss of ions formed from solar EUV in the inner circumstellar disk and in protoplanetary atmospheres.