Increasingly safety limits, especially for new products where detailed knowledge of their long-term effects is unavailable, are set as the result of a compromise between a reasonable estimate of the probable health risks and the commercial or political interests of those concerned with their manufacture. The compromise may be deliberately based on inadequate testing, or tests known to be obsolete or insensitive. Excellent tests may however be used which only test for short-term effects. Such results are then used as the basis for the legal definition of acceptable levels of toxicity. The legal definition is then used to provide general assurance in the absence of ability to communicate hard facts.
When a comparison is made between the toxicity levels considered acceptable in different countries, it is clear that some countries base their legal definitions on old methods of testing particular chemicals. The methods for testing may in some countries be treated as classified information. Following several well-publicized cases -- including the weedkiller 2,4,5-T and the pesticide Alar, which is used to "plump out" apples -- the UK government was put in the embarrassing position of having to discount evidence which led to a ban of the use of these chemicals in the US. In spite of the lack of an official ban, the products were withdrawn in the UK by the manufacturers and foods on which they had been used rejected by supermarkets.
In 1991, the UK raised by 10 times the level of dioxin considered unsafe in milk. At the previous tougher standard, all 27 samples taken at farms, dairies and retail outlets in the county of Derbyshire would have failed. Milk still allowed to be sold contained levels of dioxin higher those in milk banned from sale in Holland.