You report on recommendations that US biobanks should make participants in genome studies aware of incidental findings about their own DNA that might be medically relevant (Nature 483, 373 and Nature 483, 387; 2012). But assigning participants access rights to research that is not validated could be putting the cart before the horse.

Incidental findings about an individual's gene variants need to be verified analytically before being returned to the DNA donor (S. M. Wolf et al. Genet. Med. 14, 361–384; 2012). Even if a variant is accurately defined, further evidence of both clinical validity and utility are needed if its discovery is to be meaningful.

It would be irresponsible to follow the recommendation of Wolf et al. to communicate to the research subject the finding of one genotype variant when there may be hundreds of others related to the same phenotype, or when the phenotype varies with ethnic or environmental differences.

Investigations would be needed into whether evidence-based prevention or treatment is available for the particular gene finding, as well as into the reliability of genetic counselling based on complex, often ambiguous, risk information.

DNA donors should be informed that incidental findings must first be confirmed and translated into clinical application before information is communicated to them. We cannot side-step the process of proving clinical utility.