An intriguing new study published recently on Environmental Health News challenges the long-held assumption on which all regulatory toxicology testing is based, and poses new questions about what — and how much — of certain toxic substances merit “OK” exposure.

Toxicology tests are usually performed by giving subjects (usually rodents) high doses of a substance and monitoring the biological response. The assumption has long been that what these high doses do to the subject indicates what a lower dose could do. Using high doses is generally faster, more reliable, and cheaper.

But the new research indicates that with some pollutants, drugs, and natural substances, the effects of the toxic substance vary according to dose — meaning that the results of high doses might mean absolutely nothing when it comes figuring out the effects of lower levels of exposure.

Take, for example, the pair of rats featured prominently in a photo accompanying the study. While in-utero exposure to 100 parts per billion of the estrogenic drug diethylstilbestrol (DES) makes for scrawny adult mice, exposure to just one part per billion causes grotesque obesity, one study found.

Does this mean higher doses of toxics are better for us? Not likely. But it does mean that since most regulation is based on apparently false assumptions, we might have a whole bunch of health standards that are too weak.