Since the Chernobyl accident in 1986 there has been significant effort in a number of high-hazard industries (e.g. nuclear power, chemical and offshore oil and gas, medicine and air traffic management) to measure safety culture and safety climate. The measurement part of safety culture approaches is now relatively mature, along with the statistics to develop safety culture questionnaires and evaluate the responses. What is less mature is the approach of improving safety culture. Many safety culture surveys will show areas that need improvement, e.g. in trust, communications, or learning. The organisation that is the subject of such a survey then has the task of deciding what to do about the identified issues, and how to progress them. In a few cases this may be obvious, and there may be clear ‘quick wins’. But in some cases the solution may not be clear at all, and in others (e.g. in the case of a serious lack of trust between workforce and management) whatever solution is identified will be difficult and require sustained effort, sometimes over a period of several years. What companies do not need are bland recommendations such as ‘increase trust’, or ‘tell managers to encourage their subordinates’. Such recommendations on their own are superficial and will be unlikely to result in any real or sustainable culture change.