Valleys are key elements of mountainous landscapes, governing the transfer of water and sediment while recording the influence of tectonic and climatic forcing. Despite their importance, the quantitative characterization of valley morphology at large spatial scales remains challenging, particularly due to limitations in automated extraction methods. In this study, we compare two approaches for delineating valley extent from digital elevation models: a classical geometrical method based on topographic thresholds, and a novel pseudo-hydraulic method that estimates flood-prone areas using a simplified water-filling algorithm. Using 30 m resolution DEMs, we find that both perform well and that the conceptual difference between them is mainly significant in headwater areas. We then find that the main control on valley width is drainage area, consistent with previous work. The use of a wideness index allows us to identify local deviations from the width–area scaling, highlighting secondary controls such as lithological changes. New insights into valley morphology could be obtained through the large-scale systematic extraction of valleys across various tectonic and climatic settings.