Abstract Characterization of soil hydraulic behavior at large scales using traditional methods is time-consuming and very costly. Efficient and cheap means of providing hydrological models with such information are procedures based on pedotransfer functions (PTFs) that estimate soil hydraulic parameters from easily measurable or already available soil physical data. Major objectives of this study are to compare the prediction performance of some published PTFs and to improve their predictive capability by accounting for certain landscape variables, such as slope, aspect, and wetness index, for example. This additional information can be easily extracted from a digital elevation model of the area under study. While topographic attributes have shown potential for mapping soil properties over a region with higher precision and simplifying estimation of some model parameters, the challenge is also to examine whether, and to what extent, ancillary data of this kind can specifically contribute to improve the predictions of soil hydraulic characteristics. Since the most recent distributed hydrological models rely even more on an accurate representation of landscape features, improving PTFs with the inclusion of topographic variables is in line with this tendency. Statistical indices of goodness-of-fit are calculated to evaluate the effectiveness of the proposed methodology. It is shown, for example, that systematic biases in water retention predictions from an original PTF can be conveniently corrected by adding some primary or compound terrain attributes. The results confirm the role of terrain variables in assessing the spatial patterns of soil hydraulic characteristics.