Seismological models of the mantle are routinely developed using a range of techniques applied to different data types. For quite some time, it has been recognised that on long wavelengths models of shear-wave velocity variations show a large degree of consistency. More recently, the same has been suggested for models that describe compressional-wave velocity variations. However, controversy remains regarding models of lower mantle density variations, which provide important constraints on the nature of mantle structures, e.g. whether they are caused by thermal variations or whether additional chemical heterogeneity is required. The imaging of density structure is difficult due to a small effect on seismic observables and a strong trade-off with core-mantle boundary (CMB) topography. In addition, no consistent model of CMB topography variations exists with current models differing both in amplitude and pattern. Here, I review models of lower mantle density structure and core-mantle boundary topography from the literature, with the aim to identify which structures are consistent and what we can already learn from these models. In addition, I discuss ways in which differences between existing models may be resolved in future.