Mindful of the ethical minefield that the concept of 'race' presents, geneticists are often at great pains in their public statements to emphasize the trivial nature of genetic differences among humans. However, the high profile of endeavours such as the Human Genome Diversity Project and the HapMap is not just attributable to the potentially controversial nature of such studies. No matter how much their significance is downplayed, the implications of genetic differences between populations are important, particularly with respect to mapping genes that underlie complex traits.

So, can humans be divided into sub-populations on the basis of genetic criteria, and are such groups any more useful or valid than racial categories? In their Review (page 598), Michael Bamshad and colleagues highlight that genetic variation among humans, although limited, is highly structured and can be used to define groups that often differ from those based on phenotypic characteristics. Racial or ethnic categories can be poor proxies for genetic groupings and so can reduce the chance of detecting susceptibility genes and making correct clinical predictions.

Regardless of their scientific validity, race and ethnicity will continue to matter in a social context because such labels are important to the way individuals perceive themselves. Perhaps the best illustration of this is the increasing number of companies that offer personalized genetic history (PGH) products. On page 611, Mark Shriver and Rick Kittles point out that the ethical and statistical issues that confront the 'embryonic' PGH industry are yet to be fully addressed.

This month's issue also contains Sir David Weatherall's article on the history of thalassaemia research. This is the first article in our new series on the research history of genetic diseases, all of which will ultimately be collected together online.