The origin of the Earth and its Moon has been the focus of an enormous body of research. In this paper I review some of the current models of terrestrial planet accretion, and discuss assumptions common to most works that may require re-examination. Density-wave interactions between growing planets and the gas nebula may help to explain the current near-circular orbits of the Earth and Venus, and may result in large-scale radial migration of proto-planetary embryos. Migration would weaken the link between the present locations of the planets and the original provenance of the material that formed them. Fragmentation can potentially lead to faster accretion and could also damp final planet orbital eccentricities. The Moon-forming impact is believed to be the final major event in the Earth's accretion. Successful simulations of lunar-forming impacts involve a differentiated impactor containing between 0.1 and 0.2 Earth masses, an impact angle near 45° and an impact speed within 10 per cent of the Earth's escape velocity. All successful impacts—with or without pre-impact rotation—imply that the Moon formed primarily from material originating from the impactor rather than from the proto-Earth. This must ultimately be reconciled with compositional similarities between the Earth and the Moon.