Concern for the environment has resulted in greater scrutiny of both old and new plant protection products and increased eforts have been directed to developing more rigorous but more realistic procedures for the ecotoxicological risk characterization of these agrochemicals. These techniques include probabilistic analysis of toxicity and exposure data and better understanding of the relationship between structure and function in populations of wildlife and the role of keystone species in maintaining ecosystem functioning. The ecological risk assessment method described here is centered on the use of probabilistic distribution functions that independently describe exposure concentrations and toxicological responses of organisms to the chemical of concern. The distributions are transformed to permit calculation of linear regression parameters. The regression parameters for the two distributions are then used to determine joint probabilities which interrelate the exposure and toxicology data. For ease of presentation the results are presentedasanexceedenceplotwhichdepicts,basedontheexposuredatathepercentofspecieslikelytobea!ectedandthepercentof observations likely to cause this level of e!ect. In this paper, the use of the method is illustrated using data for chlorpyrifos in North American aquatic environments. These probabilistic risk assessment methods are being assessed for incorporation into assessment procedures in a number of regulatory jurisdictions. ( 2000 Elsevier Science Ltd. All rights reserved.