Climate is a primary control on the chemical composition of paleosols, making them a potentially extensive archive applicable to problems ranging from paleoclimate reconstruction to paleoaltimetry. However, the development of an effective, widely applicable paleosol temperature proxy has remained elusive. This is attributable to the fact that various soil orders behave differently due to their respective physical and chemical properties. Therefore, by focusing on an individual order or a subset of the 12 soil orders whose members exhibit similar process behavior, a better-constrained paleothermometer can be constructed. Soil chemistry data were compiled for 158 modern soils in order to derive a new paleosol paleothermometry relationship between mean annual temperature and a paleosol weathering index (PWI) that is based on the relative loss of major cations (Na, Mg, K, Ca) from soil B horizons. The new paleothermometer can be applied to clay-rich paleosols that originally formed under forest vegetation, including Inceptisols, Alfisols, and Ultisols, and halves the uncertainty relative to previous approaches. A case study using Cenozoic paleosols from Oregon (United States) shows that paleotemperatures produced with this new proxy compare favorably with paleobotanical temperature estimates. Global climatic events are also evident in the Oregon paleosol record, including a 2.8 °C drop across the Eocene-Oligocene transition comparable to marine records, and a Neogene peak temperature during the Mid-Miocene Climatic Optimum.