Abstract
On this basis of recent decades' research this paper presents the problems a road designer meets when he, among other goals, tries to optimise safety. The author argues that road design is only one of the factors affecting accident risk and that road users' possibilities and limitations must play a more central role in road design. Furthermore, he finds that present systems around the world are inert or conservative - the existence of guidelines for road design together with problems in realising and adopt improvements lead to a status quo of non-action. A remedy would be to create a "learning organisation". A number of possibilities for this is presented, among them what behavioural scientists could do for road design.