We exposed 34 species of stream macroinvertebrates, representing 29 families, to a 90 min pulse of the pyrethroid λ-cyhalothrin. For 28 of these species, no pyrethroid ecotoxicity data exist. We recorded mortality rates 6 days post-exposure, and the behavioral response to pyrethroid exposure was recorded using automated video tracking. Most arthropod species showed mortality responses to the exposure concentrations (0.01-10 μg L(-1)), whereas nonarthropod species remained unaffected. LC50 varied by at least a factor of 1000 among arthropod species, even within the same family. This variation could not be predicted using ecotoxicity data from closely related species, nor using species-specific indicator values from traditional ecological quality indices. Moreover, LC50 was not significantly correlated to effect thresholds for behavioral responses. Importantly, however, the measured surface area-weight ratio and the preference for coarse substrates significantly influenced the LC50 for arthropod species, with the combination of small individuals and strong preference for coarse substrates indicating higher pyrethroid sensitivity. Our study highlights that existing pesticide ecotoxicity data should be extrapolated to untested species with caution and that actual body size (not maximum potential body size, as is usually available in traits databases) and habitat preference are central parameters determining species sensitivities to pyrethroids.