It is increasingly common for water quality guidelines and risk assessments to consider the proportion of species at risk from a particular toxicant, based on the species sensitivity distribution (SSD) for that toxicant. There is a premise that the sensitivity data from species included in the SSD are sufficient to predict the effect on species for which there are no data. We discuss and review assumptions that follow this premise and find that for most toxicant SSDs include too few species, and that component species are biased toward particular taxonomic groups, common species and species from North America and western Europe. Consequently, protecting a given percentage, for example, 95%, of species in an SSD will likely protect more or less than 95% of species in nature, by an unknown amount. For the assumptions of SSDs to be better met, there is a need for tolerance data on more species, from more taxonomic and other groups, including rare species and those from widespread localities. In order to achieve this, we argue for the inclusion of rapid tests, which we define as toxicity tests designed to require less effort to conduct, relative to traditional tests, so sensitivity can be quickly and approximately determine in many species. Their use will allow for more species, more representative of natural communities, to be tested and therefore allow the construction of less biased SSDs and thus more accurate guidelines and assessments of risk.