Death would occur if sufficient quantities of any substance were taken into the body. For example, if a large group of people with similar characteristics ate half a pound of table salt, half of them would probably die. Through experiments, scientists try to establish the particular dosage of chemical (in mass per kilogram of body weight) that will result in the death of half the test animals: that is the Lethal Dose for 50% or LD₅₀. They also try to establish the point at the other end of the curve at which there is no observable effect from the substance on the animal. This is called the NOAEL: No Observable Adverse Effect Level.

Once the LD₅₀ for a substance has been established by repeated experiments with animals, it must be extrapolated to determine what the LD₅₀ would be for humans. This means adjusting the results to apply to human body weight and similar characteristics. But a toxic substance often has different effects on different species, so tests on animals cannot predict the exact effect that the substance will have on a human population. As a result, scientists are usually quite conservative in their estimates, which mean that they assume that the smallest dose that causes an effect in animals will also cause an effect in humans. In addition, scientists study the effect of a substance on human populations wherever statistics are available.

A toxic substance will sometimes combine with another substance to create a new chemical. The potential for harm of this new chemical can be greater or lesser than that of its individual components.

Another uncertainty associated with the LD₅₀ concept is that most LD₅₀ data is gained from acute exposure (single dose) testing rather than by chronic exposure. Extrapolation from these studies is complicated by the fact that chemicals are sometimes distributed differently in the body when the exposure is chronic; for example, a different target organ may be attacked, or the material may be excreted more easily.

Given these uncertainties, it is understandable why there is often considerable debate about what constitutes a “safe” level of exposure. For most substances, agency experts extrapolate conservatively from the NOAEL to set exposure limits for humans. OSHA uses Permissible Exposure Limits (PELs), while the American Conference of Governmental Industrial Hygienist (ACGIH) uses Threshold Limit Values (TLVs), to define the workroom air concentration that is considered a safe upper limit of exposure. For carcinogens and mutagens, however, there is considered to be no such “safe” exposure limit for regulatory purposes. Every exposure carries some risk.