There has been controversy about the usefulness of stratification by environmentally determined subgroups in genetic research. In cancer epidemiology only a few genetic polymorphisms cause a substantial change in risk and many genes involved in the metabolism of mutagens have little or no effect. But the inclusion of environmental measures in genetic studies might reveal environment-gene interactions. Such combined studies have been questioned, however, on the grounds that they are unlikely to increase the power of the genetic study very much. …