Variable source areas (VSAs) are hot spots of hydrological (saturation-excess runoff) and biogeochemical processes (e.g. nitrogen, phosphorus, organic carbon cycling) in the landscapes of the northeastern U.S. Despite the substantial research conducted in the past 50 years, there is still process understanding to be gained on how VSA connect with the surrounding area, how this interaction influences surface and subsurface runoff generation and chemical transport and how these processes can be captured in ungaged basins using watershed models. To determine the controls on VSA formation and connectivity, a 0.5 ha hillslope was instrumented (trenched) in the southern tier of New York, U.S. Water flux from different soil layers in the trench and upslope water table dynamics were recorded for 16 events and isotopic and geochemical tracers were measured during five events. In conjunction with the surface and bedrock topography these measurements allowed detailed characterization of the subsurface storm flow response within the VSA. Analysis revealed that the most important control on storm flow response was antecedent moisture. During events with dry antecedent conditions subsurface flow was dominated by percolation through the fragipan (i.e. cracks and macropores). Flow from below the fragipan showed a constant flow rate (0.8 mm/h), which was independent of storm size and antecedent moisture. Under wet antecedent conditions hydrological connectivity increased and subsurface flow is dominated by lateral flow through the soil atop the fragipan. During these events flow contributing slope length to the trench was five to tenfold increased. Thus, pollutant and nutrient transport from a greater distance has to be considered in water management during events with wet antecedent conditions. Application of the empirical Soil Conservation Service Curve Number method showed that discharge volumes were generally well predicted but revealed that for continuous predictions of VSA dynamics more conceptually coherent solutions need to be developed that consider the effect of antecedent moisture on runoff generation. This research shows that indirect indicators such as the average water table depth, the base flow rate prior to events or water balance estimates of the soil water content can be incorporated into watershed models to improve predictions.