Although the application of UK non-agricultural pesticides (mainly herbicides) comprises only 3% of the total amount used, similar concentrations of agriculturally and non-agriculturallyderived pesticides are routinely detected in surface waters. This has led to concern regarding the contamination of drinking water resources at concentrations above the statutory limits of the EC Drinking Water Directive (ECDWD), and the consequent risk to human health. Before the risks to drinking water resources can be fully assessed, it is important to understand and subsequently predict the chronic and transient levels of herbicide occurrence in receiving surface waters as a result of their normal application. The factors which influence herbicide transport to the aquatic environment from sites of application, particularly from the wide variety of application substrates, are not fully understood. This project addresses this lack of knowledge through an eighteen-month programme (January 1992-March 1993) of storm event herbicide monitoring on a mixed land use catchment at North Weald (Essex) which periodically received applications of common agricultural and non-agricultural herbicides including chlorotoluron, isoproturon, diuron, simazine and atrazine. To support the field monitoring programme a robust multi-residue pesticide method was developed for the simultaneous determination of the previously mentioned compounds from storm water. This was based on liquid-liquid extraction into dichloromethane and high performance liquid chromatography using photo diode array detection. The pesticide runoff data from agricultural land agreed with similar experiments carried out in the UK. The ECDWD was frequently exceeded in baseflow conditions and more frequently during storm event periods. The extent of the exceedance was found to be related to the period which had elapsed between the herbicide application and the timing of the surface water sampling. The range of application losses for the agricultural data-set was 4.0xlO-4-O.204% (median; 4.6x10-2%). The range of peak storm event concentrations was 0.03-10.0jJg/1 (median; 0.34pg/I). Similar exceedances of the ECDWD were observed during storm and non-storm conditions for discharged waters from the urban land area of the catchment. For the urban runoff data-set, the range of application losses was 0.01-45.1% (median; 0.28%) and the range of peak storm event concentrations was 0.2-238.4pg/1 (median; 0.7pg/l). The results of the monitoring programme show that the underlying factor that differentiated between the fates of herbicides applied to the North Weald catchment was the difference in the application substrate properties. Specifically, the hard surfaces, where low infiltration capacity promotes the generation of relatively high volumes of surface runoff and where poor retention behaviour exists, allow applied herbicides to be readily transported in storm event runoff to receiving surface waters. The simazine, isoproturon, chlorotoluron and diuron runoff data produced during the monitoring programme were successfully modelled using the fugacity-based Soilfug model. In the case of chlorotoluron, this model s performance was compared with a statistical model produced using multiple linear regression analysis, which showed the former approach to be superior since it required less input data and was not site specific.