Public confidence has already been undermined in the National Health Service's care.data programme in England, which will integrate medical records into a database for health-care and research purposes (Nature 507, 7; 2014). We suggest that the UK government's decision to limit release of the data should be the start of a process to develop public trust in large-scale health research, not an end to it.

Care.data first needs to provide more information to the public about how medical records will be used. Openness and clarity are essential. For example, the UK Health and Social Care Information Centre (which manages the data) can currently share pseudonymized data with 'such persons and in such form and manner ... as it considers appropriate': these choices should be fully explained to the public.

A proposed amendment to the Health and Social Care Act 2012 might narrow this discretion so that data can be shared only for 'the promotion of health'. But even this is open to wide interpretation.

Formal regulation is only one way to reassure the public. Another is to use online interfaces that allow research participants to access information and to set consent preferences (J. Kaye et al. Eur. J. Hum. Genet., in the press).