We have quantitatively investigated the mechanisms that could explain the warm, equable climate that is believed to have been typical of the mid-Cretaceous (100 million years ago). By performing simulations with a climate model based on zonal energy balance, we demonstrate that past changes in geography were important in bringing about climatic change. However, the meridional distribution of Cretaceous temperatures cannot be successfully simulated unless additional physical "feedback mechanisms" are included in the model. These mechanisms may involve cloud and meridional heat transport changes. We also conclude that paleoclimatologists should reexamine carefully both existing data and their interpretations with regard to reconstruction of Cretaceous tropical and polar surface temperatures.