A new method is presented to determine retrospectively proportional changes of species composition in a community at risk from particular concentrations of chemical stressors. The method makes estimates with some similarities to those claimed by species sensitivity distributions (SSDs) but is based on species presence/absence field data and requires assumptions that are more likely to be met. The method uses Jaccard’s index (JI), the proportion of species in common to two samples. At a similar level of contamination, the occurrence of species is usually highly variable, and thus JI values between individual pairs of samples can be low. However, by pooling samples with a similar contamination level, an increasingly complete set of species present at this level of contamination is gained. Our method involves calculating JI between randomly selected groups of samples (pooled sample sets) with similar and different levels of contamination. It then relates changes in JI to the difference in contamination and produces estimates of the proportional change in species between preselected categories of contamination. The application of the method is illustrated by using data on riverine freshwater macroinvertebrates exposed to salinity in southeastern Australia; pesticide runoff potential in the Aller River Catchment, Germany; and metal pollution (principle Cu) in the Clark Fork River Catchment, Montana, USA.