Streams in agricultural regions are severely affected by inputs from the surroundings, and pesticides in particular, act as stressors for the aquatic community. The research presented here is part of the program of the Limnology and Ecotoxicology division of the Zoological Institute of the Technical University of Braunschweig. The thesis incorporates the three most important aspects of ecotoxicological field studies. First, there is an examination of the entry routes for pesticides, and a new sampling device is presented. Next, an example for the ecological effects of contamination and the reaction of the aquatic community is given. And finally, a synthesis is achieved by constructing a biological indicator system: the extensive data analysis is transferred into the knowledge base of an expert system, making complex ecological relationships generally accessible.Section I ? III: In an agricultural catchment area in Germany I compared the pesticide contamination of entry routes. In the farmyard runoff high herbicide concentrations were found, presumably caused by cleaning the spraying equipment. The field runoff and the rainwater sewer contained less load, but included insecticides. I developed a sampling device to monitor the quality of periodically inflowing water from point sources. It inexpensively and easily enables qualitative monitoring of these entry routes. In one stream the sewage plant caused a slight but continuous contamination by herbicides, and in the other stream non-point sources caused high peaks of herbicides and contamination by insecticides.Section IV: In the New Territories of Hong Kong, China I investigated three small streams. In each stream the benthic macroinvertebrate fauna of one site upstream of an area of agricultural land was compared with a second site immediately downstream. Samples were taken at the end of the dry season (March 2000) and again (April 2000) just after heavy rainfall had caused runoff from the fields. The potential acute toxic effect of runoff became clear by focusing on the most sensitive benthic fauna. All streams showed a significant downstream decrease in the number of sensitive taxa in April, while in two streams the number of relatively tolerant taxa increased. The effect magnitude varied, which may reflect differences in the composition of the agricultural runoff.Section V ? VIII: I developed an expert system (LIMPACT) to estimate the pesticide contamination of streams using macroinvertebrate indicators. The database consisted of 157 investigations of small headwater streams with an agricultural catchment area. The pesticide load was categorised, on the basis of standardised toxicity?s, as Not Detected (n=55), Low (34), Moderate (42) and High (26) contamination. Additionally, nine water-quality and morphological parameters were evaluated with regard to their influence on the fauna and when applying LIMPACT are used to exclude unsuitable streams. The benthic macroinvertebrate fauna data were divided into four time frames (March/April; May/June; July/August; September/October) and analysed regarding the abundance of the 39 most common taxa. I differentiated between positive indicator (PI) taxa, which indicate contamination by high abundance values, and negative indicator (NI) taxa, a high abundance of which rules out contamination and indicates an uncontaminated site. The heuristic knowledge base was developed with the shell-kit D3 and contains 921 diagnostic rules. The correct diagnosis for the 157 investigations per stream and year is established by LIMPACT in 66.7 to 85.5% of the cases. The potential application of LIMPACT could be a yearly monitoring of streams and would reduce chemical analysis to the mandatory cases.