Two-component hydrograph separation was performed on 19 low-to-moderate intensity rainfall events in a 4.1-km2 urban watershed to infer the relative and absolute contribution of surface runoff (e.g. new water) to stormflow generation between 2001 and 2003. The electrical conductivity (EC) of water was used as a continuous and inexpensive tracer, with order of magnitude differences in precipitation (12-46 ??S/cm) and pre-event streamwater EC values (520-1297 ??S/cm). While new water accounted for most of the increased discharge during storms (61-117%), the contribution of new water to total discharge during events was typically lower (18-78%) and negatively correlated with antecedent stream discharge (r2 = 0??55, p < 0??01). The amount of new water was positively correlated with total rainfall (r2 = 0??77), but hydrograph separation results suggest that less than half (9-46%) of the total rainfall on impervious surfaces is rapidly routed to the stream channel as new water. Comparison of hydrograph separation results using non-conservative tracers (EC and Si) and a conservative isotopic tracer (??D) for two events showed similar results and highlighted the potential application of EC as an inexpensive, high frequency tracer for hydrograph separation studies in urban catchments. The use of a simple tracer-based approach may help hydrologists and watershed managers to better understand impervious surface runoff, stormflow generation and non-point-source pollutant loading to urban streams. Copyright ?? 2007 John Wiley & Sons, Ltd.