Transportation Research Record: Journal of the Transportation Research Board, No. 1897, TRB, National Research Council, Washington, D.C., 2004, pp. 142–147. 142 Monthly time series data related to road traffic and pedestrian casual- ties and fatalities in Great Britain are analyzed and trends are identi- fied. These forecasts are based on extrapolation of the absolute number of casualties and not the ratio of casualties to 100 million vehicle kilome- ters. A short review of influences on these trends is presented, and a number of statistical forecasting autoregressive integrated moving aver- age models are then constructed. Predictions produced are compared with government targets, and it is found that targets for fatal, serious, and slight casualty reduction are on track to be met. However, there are fewer grounds for optimism regarding pedestrian casualties. A key performance measure of a nation’s transport system is the number of people who are killed or seriously injured. National gov- ernments provide targets that traffic managers, infrastructure design- ers, vehicle manufacturers, and the legal system strive to achieve. The latest set of targets for Great Britain are for 2010, by which year it is hoped that compared to the average for 1994 to 1998, the num- ber of fatal or seriously injured in road accidents will be reduced by 40%, the number of children killed or seriously injured by 50%, and the number of slight injuries per 100 million vehicle kilometers by 10%. The previous set of targets, set in 1987, sought to reduce deaths and serious injury by one-third by the year 2000, compared to the average of 1981 to 1985. This was achieved; road deaths fell by 39% and serious injuries by 45%. The success in Great Britain has come about through legislative changes intended to alter driver behavior, improve infrastructure, and improve crashworthiness of vehicles. A chronology of events affecting road safety was published by the Department of Transport (1). The number of casualties is of concern throughout Europe, where there are more than 40,000 deaths and 1,700,000 injuries per year, costing some 160 billion €, and the young are most affected (2). Considering the ranking of European nations for number of fatali- ties per 1 billion km driven in 2001, Great Britain compares favorably with its European neighbors and other Organisation for Economic Co-operation and Development countries. Its improvement in rates of road traffic injuries makes Great Britain one of the safest European countries, but its 1997 child pedestrian fatality rate per 100,000 people is one of the worst in Europe. Of con- cern is that since 1991, the number of road traffic casualties in Great Britain has shown a slight upward trend. Figure 1 shows that the accident severity trends are somewhat dif- ferent from the total. Thus the numbers of fatal and severe casual- ties have decreased markedly over the period, whereas the number of slight injuries has increased. Since 1991, the number of vehicle kilo- meters traveled in Great Britain has increased by more than 15%; the risk of fatal or serious injury has substantially decreased by 24.5% and 28.1%, respectively. This paper explores recent casualty time series in Great Britain and forecasts these series to 2010. Most forecasting approaches predict casualties per 100 million km, which is problematic because it implies predicting a time series that is hard to estimate and is integral in the time series of casualties. Thus an attempt is made here for a straight- forward forecasting of casualty numbers. This will reflect the strong seasonality that is inherent in the casualty time series. It is hoped this will be useful to those involved with road safety, for determining if goals for the number of those killed or seriously injured can be met or if more effort is required. REVIEW OF PREVIOUS WORK The main approach to forecasting killed or seriously injured (KSI) casualties has been to take a time series of annual rates and fit a neg- ative exponential model and then to extrapolate from this. Sometimes allowance is made through use of disturbance terms for special events, like the introduction of legislation to make use of seat belts compul- sory, but in general the models are univariate and incorporate few explanatory variables. A good example is the work of Broughton (3), who fitted extensions of the model For fatalities, this model gave surprisingly good forecasts, pre- dicting the number of fatal casualties in 2000 to be 3,312 with a 90% prediction interval of 2,892 to 3,826; there were 3,409 fatali- ties. Broughton’s forecasts of KSI numbers and all casualties under- predicted by 4.8% and 24.8%, respectively. A major problem with this approach is that one must forecast the number of kilometers driven to allow the output of the model to be expressed as the number of casualties. The use of kilometers as a denominator is contentious because most casualty accidents occur relatively close to the place of resi- dence of the casualty (4, 5). In addition, the certainty of estimation of a nation’s annual driving is debatable. However, Nilsson argued that for comparisons over time and between nations, mortality rates should be divided by kilometerage, but Nilsson acknowledged that this is not completely satisfactory (6 ). There has been much discussion in the literature about underly- ing variables that could explain casualty trends. Smeed (7 ) and Sivak (8), for example, pointed out that many factors influence a country’s log casualties traffic volume year an intervention term ( ) = + + a b  Predicting Casualty Numbers in Great Britain Robert Raeside and David White Transport Research Institute, Napier University, Sighthill Court, Edinburgh EH10 4BN, Scotland.