Hazard assessment of hundreds of observed and potential contaminants in fish, sediment, and water of the Great Lakes is necessary to determine impact on fishery sources and other aquatic biota. The hundreds of new compunds introduced each year have few measured properties. Mathematical models based on quantitative structure-activity relationships (QSARs) provide rapid, inexpensive estimates of physical properties and toxicites. These models are useful when the contaminants are not commercially available for testing. To develop these predicitive models, acute bioassay results (48-hr EC50 of Daphnia pulex) of 62 compounds were regressed against several modes of molecular connectivity indices. The compunds were selected from the more hazardous classes in a list of 476 compunds tentatively identified by GC/MS in Great Lakes fish.