Dryland river systems support a diverse aquatic biota in semi-arid and arid landscapes, but they are increasingly vulnerable to natural and human impacts. This study provides the first modern diatom training set (n = 38) from three dryland rivers in northern Patagonia, Argentina. The relationship between surface-sediment diatoms and contemporary environmental variables was explored using multivariate analyses. The results showed that surface-water chemistry varies among rivers in accordance with local-scale factors such hydro-climatic variability and anthropic activities. A total of 378 diatom taxa were identified but only 45 taxa occurred with relative abundances higher than 2% in at least one sediment sample. The Colorado and Negro rivers were characterized by high abundances of small tychoplanktic fragilarioids whereas the Chubut River was dominated by epiphytic and planktic taxa. The canonical correspondence analyses (CCA) showed that conductivity, pH and HCO 3 − were the main environmental variables governing diatoms distributions. A transfer function was developed for water conductivity using unimodal methods because of partial CCA and λ 1 /λ 2 ratio revealed that this variable had the largest unique effect on diatom variance. A weighted averaging (WA) model with tolerance downweighting and classic deshrinking provided a reasonably robust model (r 2 jack = 0.94 and RMSEP = 0.249 log 10 μS cm −1 ). However, the diatom-inferred conductivity values should be carefully interpreted because of the uneven distribution of sampling sites and the dominance of small araphid diatoms belonging to the genera Staurosira, Pseudostaurosira and Punctastriata. By their benthic habitat, the distribution of these taxa could be influenced by other factors as light and substrate in addition to water chemistry. The autoecological information achieved through this study will provide the basis for a better insight into the hydrological responses of the dryland rivers in northern Patagonia to climate and environmental changes.