The reliability of behavioural data constitutes a major concern in the neuroscience field. Indeed, discrepancies in the behavioural patterns of mice or rats in the same anxiety tests performed in different laboratories have been reported recently. The question raised by such data addressed, in particular, the selection and breeding of two lines of rats on the basis of their high (HAB) and low (LAB) anxiety-related behaviour in the elevated plus-maze test at the Max Planck Institute of Psychiatry in Munich (Germany). As the majority of the behavioural data in these animals has been derived from research carried out in this institute, the aims of the present study were: (1) to test the reliability of the differences in anxiety-related behaviour of these rats in two other laboratories (Villeneuve d'Ascq, France and Innsbruck, Austria); and (2) to determine how the different behavioural traits were associated in both HAB and LAB rats by a principal component analysis. Results were in agreement with the studies performed in Munich, as the divergence in anxiety-related behaviour of the two lines was highly consistent in all tests performed in Villeneuve d'Ascq and Innsbruck. Moreover, the most important parameters to discriminate the two lines were similar to those found in a previous study. Finally, the principal component analysis again confirmed that the selection of HAB and LAB rats is based on anxiety-related behaviour rather than locomotor activity.