Measuring anxiety-like behaviour in mice has been mostly undertaken using a few classical animal models of anxiety such as the elevated plus-maze, the light/dark choice or the open-field tests. All these procedures are based upon the exposure of subjects to unfamiliar aversive places. Anxiety can also be elicited by a range of threats such as predator exposure. Furthermore, the concepts of 'state' and 'trait' anxiety have been proposed to differentiate anxiety that the subject experiences at a particular moment of time and that is increased by the presence of an anxiogenic stimulus, and anxiety that does not vary from moment to moment and is considered to be an 'enduring feature of an individual'. Thus, when assessing the behaviour of mice, it is necessary to increase the range of behavioural paradigms used, including animal models of 'state' and 'trait' anxiety. In the last few years, many mice with targeted mutations have been generated. Among them some have been proposed as animal models of pathological anxiety, since they display high level of anxiety-related behaviours in classical tests. However, it is important to emphasise that such mice are animal models of a single gene dysfunction, rather than models of anxiety, per se. Inbred strains of mice, such as the BALB/c line, which exhibits spontaneously elevated anxiety appear to be a more suitable model of pathological anxiety.