The similarity of tiled-drained watersheds to karst drainage basins can be used to improve understanding of watershed-scale nutrient losses from subsurface tile drainage networks. In this study, short-term variations in discharge and chemistry were examined from a tile outlet collecting subsurface tile flow from a 963 ha agricultural watershed. Study objectives were to apply analytical techniques from karst springs to tile discharge to evaluate water sources and estimate the loads of agricultural pollutants discharged from the tile with conduit, intermediate and diffuse flow regimes. A two-member mixing model using nitrate, chloride and specific conductance was used to distinguish rainwater versus groundwater inputs. Results indicated that groundwater comprised 75% of the discharge for a three-day storm period and rainwater was primarily concentrated during the hydrograph peak. A contrasting pattern of solute concentrations and export loads was observed in tile flow. During base flow periods, tile flow consisted of diffuse flow from groundwater sources and contained elevated levels of nitrate, chloride and specific conductance. During storm events, suspended solids and pollutants adhered to soil surfaces (phosphorus, ammonium and organic nitrogen) were concentrated and discharged during the rapid, conduit flow portion of the hydrograph. During a three-day period, conduit flow occurred for 5.6% of the time but accounted for 16.5% of the total flow. Nitrate and chloride were delivered primarily with diffuse flow (more than 70%), whereas 80-94% of total suspended sediment, phosphorus and ammonium were exported with conduit and intermediate flow regimes. Understanding the water sources contributing to tile drainage and the manner by which pollutant discharge occurs from these systems (conduit, intermediate or diffuse flow) may be useful for designing, implementing and evaluating non-point source reduction strategies in tile-drained landscapes. ?? 2007 Elsevier B.V. All rights reserved.