Abstract. The reconciliation of mantle chemistry with the structure of the mantle inferred from geophysics and dynamical modelling has been a long-standing problem. This paper reviews three main aspects. First, extensions and refinements of dynamical modelling and theory of mantle processing over the past decade. Second, a recent reconsideration of the implications of mantle heterogeneity for melting, melt migration, mantle differentiation and mantle segregation. Third, a recent proposed shift in the primitive chemical baseline of the mantle inferred from observations of non-chondritic 142Nd in the Earth. It seems most issues can now be resolved, except the level of heating required to maintain the mantle's thermal evolution. A reconciliation of refractory trace elements and their isotopes with the dynamical mantle, proposed and given preliminary quantification by Hofmann, White and Christensen, has been strengthened by work over the past decade. The apparent age of lead isotopes and the broad refractory-element differences among and between ocean island basalts (OIBs) and mid-ocean ridge basalts (MORBs) can now be quantitatively accounted for with some assurance. The association of the least radiogenic helium with relatively depleted sources and their location in the mantle have been enigmatic. The least radiogenic helium samples have recently been recognised as matching the proposed non-chondritic primitive mantle. It has also been proposed recently that noble gases reside in a so-called hybrid pyroxenite assemblage that is the result of melt from fusible pods reacting with surrounding refractory peridotite and refreezing. Hybrid pyroxenite that is off-axis may not remelt and erupt at MORs, so its volatile constituents would recirculate within the mantle. Hybrid pyroxenite is likely to be denser than average mantle, and thus some would tend to settle in the D" zone at the base of the mantle, along with some old subducted oceanic crust. Residence times in D" are longer, so the hybrid pyroxenite there would be less degassed. Plumes would sample both the degassed, enriched old oceanic crust and the gassy, less enriched hybrid pyroxenite and deliver them to OIBs. These findings can account quantitatively for the main He, Ne and Ar isotopic observations. It has been commonly inferred that the MORB source is strongly depleted of incompatible elements. However it has recently been argued that conventional estimates of the MORB source composition fail to take full account of mantle heterogeneity, and in particular focus on an ill-defined "depleted" mantle component while neglecting less common enriched components. Previous estimates have also been tied to the composition of peridotites, but these probably do not reflect the full complement of incompatible elements in the heterogeneous mantle. New estimates that account for enriched mantle components suggest the MORB source complement of incompatibles could be as much as 50–100 % larger than previous estimates. A major difficulty has been the inference that mass balances of incompatible trace elements could only be satisfied if there is a deep enriched layer in the mantle, but the Earth's topography precludes such a layer. The difficulty might be resolved if either the Earth is depleted relative to chondritic or the MORB source is less depleted than previous estimates. Together these factors can certainly resolve the mass balance difficulties.