The qualitative assessment of animal behaviour summarizes the different aspects of an animal's dynamic style of interaction with the environment, using descriptors such as ‘confident’, ‘nervous’, ‘calm’ or ‘excitable’. Scientists frequently use such terms in studies of animal personality and temperament, but, wary of anthropomorphism, are reluctant to do so in studies of animal welfare. We hypothesize that qualitative behaviour assessment, in describing behaviour as an expressive process, may have a stronger observational foundation than is currently recognized, and may be of use as an integrative welfare assessment tool. To test this hypothesis, we investigated the inter- and intraobserver reliability of spontanous qualitative assessments of pig, Sus scrofa, behaviour provided by nine naı̈ve observers. We used an experimental methodology called ‘free choice profiling’ (FCP), which gives observers complete freedom to choose their own descriptive terms. Data were analysed with generalized Procrustes analysis (GPA), a multivariate statistical technique associated with FCP. Observers achieved significant agreement in their assessments of pig behavioural expression in four separate tests, and could accurately repeat attributing expressive scores to individual pigs across these tests. Thus the spontaneous qualitative assessment of pig behaviour showed strong internal validity under our controlled experimental conditions. In conclusion we suggest that qualitative behaviour assessment reflects a ‘whole animal’ level of organization, which may guide the intepretation of behavioural and physiological measurements in terms of an animal's overall welfare state.