Available water capacity (AWC) is a fundamental soil hydraulic property controlling plant water supply and drought response, yet existing AWC estimates for Switzerland's forests are an order of magnitude coarser than the scale of an individual tree’s root system. Here we produce 25 m resolution maps of mineral soil AWC for Switzerland's forest soils using ten published pedotransfer functions (PTFs) applied to a common set of gridded 25 m digital soil maps. We introduce a novel method for evaluating these PTF-derived AWC estimates using early leaf discoloration, an indicator of extreme drought stress, in European beech (<em>Fagus sylvatica </em>L<em>.</em>). Performance of single feature logistic regression models, using estimated AWC as a predictive feature and September 2018 early leaf discoloration as a target, varied substantially by PTF (ROC-AUC between 0.42 and 0.63). The strongest relationships between AWC and discoloration showed a negative trend and relatively strong model calibration (R² up to 0.82). Combining AWC with total 2018 summer precipitation and maximum 2018 summer temperature had little impact on model discrimination, but improved model calibration (R² up to 0.90) in late summer 2018 with discoloration proportions of nearly 80% at the most susceptible sites. Averaging the results of the four comparatively best-performing PTFs in these models produced two 25 m soil AWC products for the Swiss forest (mean 1 m and 2 m depth AWC) with uncertainty estimates based on variation among PTFs (± 1σ). While both the mapped AWC values and the relative performance of individual PTFs remain dependent on soil properties of the underlying digital soil maps, these new AWC estimates are immediately applicable to forest drought monitoring and species distribution, ecohydrological, and land-surface modelling.