A considerable amount of psychological research has been conducted on human behaviour and its relationship to accidents. However, the research covers very wide fields and the findings drawn are so varied that at first sight they are confusing and there does not appear to be any major useful lead. I propose a discussion of a number of research findings to illustrate the variety and complexity of the human behaviour factors involved in considering safety. Such discuss- ion might point to some conclusions which could lead to an action programme to improve safety records.