Abstract
The risk homeostasis theory suggests that safety measures will not reduce accident loss unless they lower "target levels of risk". The theory is plausible but untestable- target levels of risk are unmeasurable, and no all- embracing index of accident loss exists. Risk compensation effects are frequently underestimated. If danger is defined as the potential of some thing or activity to cause harm, then accident statistics are a worthless measure of it. The "road safety community" needs a new agenda. It should be less concerned about protecting people from themselves. Probably and impossible task, and more concerned about achieving a fairer distribution of road traffic risk.