There is evidence that a sizeable proportion of adults distrust children's testimony. An analysis of individual expressions of this distrust suggests that it is based on four main ideas. These are that there are age trends in the reliability of children's testimony such that: firstly, the tendency to confabulate, or recall on the basis of what was probable rather than on what was seen, decreases with age; secondly, the tendency to confuse fact with fantasy decreases with age; thirdly, the tendency uncritically to accept misinformation about a witnessed event after the event decreases with age, and finally, susceptibility to social pressures which may distort testimony decreases with age. The experiments reported in this thesis were designed to test these hypotheses. Only the final hypothesis was supported in its simplest form, and even this hypothesis was not supported if subjects had already committed themselves to an account of the details of an event, prior to being exposed to social pressures on these details. These findings suggest that age is an unreliable predictor of distortions in children's event recall, and that problems with children's testimony may be specific to situations rather than to particular age groups. The results of the experiments are compared with traditional ideas about child witnesses, and the idea that it may be possible to enhance the reliability of children's testimony is considered.