Climate sensitivity (S) describes Earth’s resulting global mean surface temperature adaptation (ΔT in K) to a changing radiative forcing (ΔR, in Wm−2) at the top of the atmosphere, thus, S =ΔT/ΔR. Natural climate changes have occurred in the pre-anthropogenic past due to changes in insolation, atmospheric composition, and feedback responses within the Earth system. Using observations and modeling, paleoclimate studies quantify these changes to estimate equilibrium climate sensitivity (a critical but thus far poorly constrained system component in models for projections of future climate change). However, problems exist between studies, regarding comparability of concepts and reference system definition. These obstruct the development of a coherent picture of equilibrium climate sensitivity through time, and so of its dependence on the background climate state. We assess these issues, in particular the confusion over feedback versus forcing, and propose a pragmatic way forward