Relevant research related issues are briefly examined. The known influence of experience implies the possibility of effective education and training. At present, however, there is no sound evidence to show that effective programmes exist, to support the suspicion of a special effectiveness for motorcyclists, or to suggest that a licence test is an effective crash countermeasure - either as a selection device or as an incentive for pre-licensing education. Great expenditure on fairly ineffective measures results in loss of opportunity to benefit from more effective, mainly engineering, measures. Longer term driver behaviour research seems to have a high priority.