Foreseeable incidents that could occur at process industry sites have the potential to harm the on and/or off-site population. If an incident is severe it could affect many people causing deaths and injuries. The regulatory authorities in many countries require a Safety Report to be prepared by operators of certain sites. Information to be reported can include an assessment of the extent and severity of the consequences of identified major accidents. The outcome of this assessment should inform the operator's appraisal of the overall risks, decisions on risk reduction measures and emergency planning. Modelling the consequences of an incident, for example with one of the numerous software packages available, can help to establish the extent of a foreseeable major accident scenario. Estimation of the severity, in terms of potential (non-fatal) injuries and deaths, requires the setting of harm criteria. Harm criteria are not highly reliable. Harm criteria associated with fatal injury are usually the best developed for practical reasons, not least of which is that the harm is well-defined. Generally, the uncertainty in the process of estimating the number of people who could be harmed becomes larger as the severity of the harm being considered becomes lower. An obvious question is whether it is better to estimate numbers of potential deaths and injuries using separate harm criteria, or to focus on estimating the number of possible deaths and then to estimate the number of injuries by use of a ratio. This begs the question of whether a suitable ratio can be derived. Here, a study of historical data has been used to examine whether the relationship between the number of deaths resulting from an incident and the number of injuries reported can be described by a simple ratio. No comparable earlier work has been identified.