An original species-selection method for the building of test batteries is presented. This method is based on the statistical analysis of the biological and ecological trait patterns of species. It has been applied to build a macroinvertebrate test battery for the assessment of sediment toxicity, which efficiently describes the diversity of benthic macroinvertebrate biological responses to toxicants in a large European lowland river. First, 109 potential representatives of benthic communities of European lowland rivers were selected from a list of 479 taxa, considering 11 biological traits accounting for the main routes of exposure to a sediment-bound toxicant and eight ecological traits providing an adequate description of habitat characteristics used by the taxa. Second, their biological and ecological trait patterns were compared using coinertia analysis. This comparison allowed the clustering of taxa into groups of organisms that exhibited similar life-history characteristics, physiological and behavioral features, and similar habitat use. Groups exhibited various sizes (7-35 taxa), taxonomic compositions, and biological and ecological features. Main differences among group characteristics concerned morphology, substrate preferendum and habitat utilization, nutritional features, maximal size, and life-history strategy. Third, the best representatives of the mean biological and ecological characteristics of each group were included in the test battery. The final selection was composed of Chironomus riparius (Insecta: Diptera), Branchiura sowerbyi (Oligochaeta: Tubificidae), Lumbriculus variegatus (Oligochaeta: Lumbriculidae), Valvata piscinalis (Gastropoda: Valvatidae), and Sericostoma personatum (Trichoptera: Sericostomatidae). This approach permitted the biological and ecological variety of the battery to be maximized. Because biological and ecological traits of taxa determine species sensitivity, such maximization should permit the battery to better account for the sensitivity range within a community.