In infancy and early childhood ‘shyness’ may be viewed as one aspect of the widely studied concept of ‘behavioural inhibition’ – defined as a child’s initial withdrawal to unfamiliar or challenging events (e.g., Kagan, 1989; 1994 and this volume, Chapter 2) – an aspect in which the events are restricted to social stimuli. With increasing age, the term ‘shyness’ has been applied not only to observed behaviour, but also to inner feelings. For example, Jones, Briggs and Smith (1986: 630) define shyness as ‘a tendency to respond with heightened anxiety, self-consciousness, and reticence in a variety of social contexts; a person high in the trait of shyness will experience greater arousal than a person low in shyness independent of the level of interpersonal threat in the situation’. Now with young children, shyness may be assessed by means of questionnaires to parents or teachers, and results from these informants do indeed correlate significantly with each other and with direct observations (e.g., Stevenson-Hinde and Glover, 1996). However, in addition to the obvious point that questionnaires carry the risk of cross-temporal assessments being biased by relying on the same observers, we have found that mothers of securely attached children tend to over-estimate their children’s shyness as compared with our observations, while mothers of insecurely attached children do the opposite (Stevenson-Hinde and Shouldice, 1990; see also Sameroff, Seifer and Elias, 1982; Stevenson-Hinde and Shouldice, 1995a; Vaughn, Taraldson, Crichton and Egeland, 1981). Since attachment is our other main interest, it was essential to avoid this sort of respondent-error in our assessments of shyness. Thus, in our own work, we tend to use questionnaires only as a ‘back-up’ (e.g., for screening purposes, see Stevenson-Hinde and Glover, 1996). Following Kagan’s lead, we rely primarily on direct observations of behaviour. Thus direct observations will be the focus of the present chapter, with assessments of behavioural inhibition from the Madingley studies involving responses to social stimuli only, and behavioural inhibition from the other studies cited involving Kagan’s more general definition.