Safety rules are a universal phenomenon of any technological activity. Newcomers are taught them from rules books, or are expected to pick them up from working with experienced operators. In complex and dangerous organisations they can form many volumes of a safety manual, with procedures running into scores and applying to operations, maintenance, design, selection, training, purchasing and storage. When accidents happen, one of the first recourses of the investigation team is commonly to find out if safety rules have been broken. If they have, there is a prima facie case for allocating responsibility and liability, and often also for focusing change to improve prevention. The response to tragic accidents in activities which have not, up to now, been subject to any regulation, is often to formulate rules for them too, so that dangerous hobbies and even public pastimes such as hill walking, canoeing or swimming get to resemble more and more the traditional dangerous occupational pursuits. Safety rules are therefore seen as central to all attempts to prevent accidents and achieve safety. The meta-analysis made by Guldenmund (2000) of safety culture and climate studies showed that more than half identified safety rules and attitudes towards their use and violation as significant dimensions of safety culture (see also Hale et al 2002, Swuste & Guldenmund 2002 for more recent studies confirming this importance). Yet it is surprising how little literature there is about how to manage safety rules effectively, how to decide what rules are needed, how to prepare and formulate them and how to promulgate them and ensure they stay appropriate.