The existing models for estimating cation exchange capacity (CEC) from easy‐to‐measure hygroscopic water content (θh) were based on a single water activity (aw) value rather than on the processes that govern soil water vapor adsorption for a distinct aw range. Here, we present a new CEC estimation model based on θh data of 119 soils with different clay mineralogy (i.e., illitic [IL], montmorillonitic [ML], and kaolinitic [KA] samples) and organic carbon (OC) contents for the aw range from 0.23 to 0.57 (Δθ0.23–0.57) and validate its performance. Based on the hypothesis that multilayer adsorption exhibits a higher correlation with CEC than monolayer adsorption and capillary condensation, the aw range from 0.23 to 0.57 was chosen with CEC calculated as CEC = k × Δθ0.23–0.57. The performance of the new model is compared to the Arthur (2017) model and the Torrent (2015) model, which considers a single θh value. The average proportionality coefficient (k) varied with the dominant clay mineralogy of the investigated soils. For soils dominated by 2:1 clay minerals (i.e., IL and ML), the new model showed a good estimation accuracy (Nash‐Sutcliffe model efficiency [E] ≥ 0.85; root mean squared error [RMSE] ≤ 4.18 cmol(+) kg−1). The new model performed better for IL and ML samples than for KA samples, and yielded more accurate CEC estimations than the Arthur model and Torrent model for soils with 2:1 clay minerals. For soil with high OC content (>23.2 g kg−1), the new model slightly underestimated CEC (E = 0.66; RMSE = 5.87).