Health and safety regulations have always been concerned with risk, though not always overtly. The quantitative expression of risk does not appear in regulations and rarely in guidance materials but is inherent in the policy underlying the development of regulations and in their practical application. The ways in which actual and perceived expressions of risk are used in regulatory actions differ widely. Some dangers are treated as unacceptable and the regulatory policy is to exclude them totally. In the real world, such policies are never completely successful. The head-on collision of trains in main-line working is one example. The deliberate use of known carcinogens as pesticides is another. Other dangers are recognized as inevitable but as being reducible in degree. The regulatory activity is then aimed at limiting the extent to which a citizen can expose other citizens to this danger and, more recently, the extent to which he is permitted to put himself at risk. The balance of risks and benefits and of one risk with another underlie decisions in these cases. Not only consequences but probabilities become relevant. In the past, all, and even now most, of the regulation of risk has been on a non-quantitative basis. Increasingly, there is a desire to make the process more quantitative and to introduce the idea of acceptability. This change is provided for by many of the features of the Health and Safety at Work Act of 1974. Within the framework of that Act, the Health and Safety Commission and its operating arm, the Health and Safety Executive, are developing the more systematic use of quantitative methods of controlling hazards from work activities. The regulation of risk is a growth industry and it behoves us all to clarify our objectives. An aim of zero risk would not be to the benefit of society, but its replacement by more suitable aims is a long and complicated process.