Figure 1 shows fatality rates for selected modes for a time period of two decades or more. It is clear that safety in most modes has improved over the last 25 years. However, for several of the modes, the greatest improvement in fatality rates tended to occur in the earlier years of the period.

The activity measures used as denominators are not the same for all modes. For highway travel, exposure to risk is approximately proportional to distance traveled, hence the use of vehicle-miles as the denominator. For aviation, the greatest proportion of crashes occurs during takeoff and landing; hence risk is approximately proportional to the number of operations (measured as departures). Data on departures are not available for general aviation for recent years, so hours flown is used instead. For some means of travel, there are no good measures of the risks entailed. For example, while over 4,900 pedestrians were struck by motor vehicles and died in 1999, exposure measures are lacking because good data are not available for the amount of time, distances, or other circumstances of pedestrian travel.