[en] A comparison of hazards is based on an attempt to quantify risks. In the case of exposure to radiation, dose is used as a surrogate for risk, with risks based on an extrapolation of stochastic effects observed at doses well above occupational limits. In comparing risks of fatalities, it is worth keeping in mind that the risk expression arising from uniform exposure to radiation throughout a working lifetime peaks at late ages, and for the same fatality risk the years of life lost from exposure to radiation are much less than for prompt fatalities, for example, from industrial accidents. In considering the topic, a first question is, how do we judge the adequacy of safety standards for avoidance of hazards? There are three widely used approaches. One of these, used commonly in the case of exposure to workplace chemicals, is to establish or estimate a threshold (e.g. for the concentration in air for continuous breathing) below which no harm is likely and accept exposures up to some fraction of the threshold as 'safe'. In the case of exposure to radiofrequency radiation, thresholds have been established for thermal effects and safety factors applied to these to derive occupational exposure limits. A second approach is used in the case of some genotoxins and asbestos for which there may be no threshold, and it is then necessary to establish an exposure-effect relationship and reduce exposures to a level where risks are acceptable or at least tolerable. The question then becomes, what is a tolerable risk? The third approach to standards setting is directed more particularly at avoidance of accidents which may result in 'grievous bodily harm' (e.g. a containment failure for liquefied chlorine). In principle, such events can be avoided by adopting appropriate engineering and administrative systems, and the probability of occurrence can be assessed through fault tree or probabilistic safety analysis. This allows some estimate of the risk of such an event to be made, although not necessarily an accurate estimate of the risk of health consequences. 'Safety' is judged acceptable if the risk is below some value which is considered attainable and acceptable for the industry