The European Union's (EU) Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) directive, which took effect a year ago, on 1 June 2007, is widely regarded as the strictest chemical safety law in the world. Unlike the 1981 EU legislation it replaces, or the US Toxic Substances Control Act, REACH applies to all existing chemicals, not just new ones.

If nothing else, that requirement is likely to cause six months of immense stress at the organization created to administer REACH, the European Chemicals Agency in Helsinki. The agency will open its doors to registration on 1 June this week. And with every chemical manufacturer or importer in Europe facing a 1 December deadline to do the initial registration on every compound it had put on the market before 1981, agency staffers are braced for an estimated 180,000 applications.

More importantly, however, the requirement to test all those old chemicals has made REACH a case study in difficult choices. In following one policy — ensuring the safety of its citizens — the European Union may have severely strained another: lessening the use of animals in regulatory toxicology, and in research in general. Working through that backlog without sacrificing a lot of animals will be impossible.

Happily, the actual number of sacrifices is likely be less dramatic than it might have been. Partly this is because safety data are already thought to exist on many of the older chemicals (although no one has comprehensive statistics). And partly it is because of REACH itself. In 2002, while the directive was still being drafted, the European Union expanded and accelerated work at the European Centre for the Validation of Alternative Methods: a small laboratory in Ispra, Italy, devoted to finding kinder and more-refined toxicity tests that would require fewer, or no, animals. That work has begun to pay off. For example, the notorious LD50 test is required, which identifies the dose that kills half of the tested animals, now involves around ten rats rather than 45. And instead of looking for a dose that causes outright (and painful) skin allergy in guinea pigs, toxicologists can now monitor mice for the painless swelling of lymph nodes, the first sign of allergy.

These are the only two animal tests required for chemicals produced in relatively small quantities (1–10 tonnes a year). Those produced in smaller quantities require no animal testing at all. Not so happily, however, the refinements will have little effect when it comes to chemicals produced in larger volumes. REACH will require many more tests on such compounds, including, in some cases, their effects on reproduction and fertility, which use 3,000 animals per chemical.

Beyond REACH, moreover, the 2003 amendment to the EU Cosmetics Directive will soon add even more time pressure: an end to acute toxicity testing of cosmetics (including soaps and shampoos) in animals by 2009, and the longer and more involved repeat-dose testing by 2013. Scientists consider the latter deadline to be hopelessly unachievable. There is simply no way in the foreseeable future that they will be able to model and predict all the ways that a foreign molecule might interact over time with the hundreds of types of tissues in the human body.

Yet that is all the more reason to put some serious effort into tackling the problem. Revolutionizing regulatory toxicology won't be cheap and it won't be quick, although systems biology — the integration of genomics, proteomics and all the other '–omics' — could prove to be a particularly powerful ally. But it should be a priority. And, as shown by the cosmetics directive's unrealistic deadline (fortunately renegotiable in 2011), scientists have to be brought into policy decisions at the beginning, and listened to carefully.