The groundfish trawl fishery operating off the coasts of Oregon and Washington 1985-1987 caught six major assemblages of species which could be treated as units in developing mixed-species management plans. Eighty-one percent of the tows observed on commercial vessels were consistently placed in one of the assemblage designations using three multivariate techniques. Two of the assemblages were dominated by a single species, pink shrimp (Pandalus jordani) or widow rockfish (Sebastes entomelas). The other assemblages identified were: a deepwater rockfish assemblage, a deepwater Dover sole assemblage, a nearshore mixed-species assemblage, and a bottom rockfish assemblage. The assemblage designations of the consistently placed tows were predicted with an estimated 85% average accuracy using discriminant functions based on the gear used and bottom depth fished. Fishermen had different targets (intended catch) for each assemblage caught. The mixed-species assemblages had several targets, representing at least some of the dominant species in the assemblage. Targeting and discarding information indicated that fishermen did not always intend to catch the species together; there were unintentional or unavoidable catches of all the major species except for shrimp in the shrimp assemblage. Discarding occurred in all the assemblages, primarily due to unmarketable species or fish that were too small to market. Monitoring the assemblages over time could be accomplished by using the defined strategies with logbook data, particularly if the large and small rockfish categories were used to consistently separate shelf and deepwater rockfish. The defined strategies could not effectively predict the research cruise assemblages or catches; research data do not accurately describe commercial catches. / Graduation date: 1994