Understanding hydrological flow pathways is important for modelling stream response, in order to address a range of environmental problems such as flood prediction, prediction of chemical loads and identification of contaminant pathways for subsequent remediation. This paper describes the use of parametrically efficient, low order models to identify the dominant modes of stream response for catchments within the Upper Eden, UK. A first order linear model adequately identified the dominant mode in all but one of the sub-catchments. A consistent pattern of time constants and pure time delays between catchments was observed over different periods of data. In the nested catchments, time constants increased as the catchment size increased from 1.1 km2 at Gais Gill (2–7 h) to 69.4 km2 at Kirkby Stephen (5–10 h) to 223.4 km2 at Great Musgrave (7–16 h) to 616.4 km2 at Temple Sowerby (11–22 h), but Blind Beck (a small catchment 8.8 km2, time constants 11–21 h) had time constants most similar to Temple Sowerby. This was attributed to a combination of the storage role of permeable rock strata, where present, and the effect of scale on sub-surface and channel routing. A first order model could not be identified for the 1.0 km2 Low Hall catchment, which comprises permeable sandstone overlain by Quaternary sediments. A second-order model of Low Hall stream showed a higher proportion of water taking a slower pathway (76% via a slow pathway; time constant 252 h) than a model with the same structure for the 8.8 km2 Blind Beck (46% via slow pathway; time constant 60 h), where only 38% of the basin was underlain by the same permeable sandstone. This highlights the need to quantify the role of deep pathways through permeable rock, where present, in addition to the effect of catchment size on response times.