Most previous studies addressing the issue of inter-relationships between health behaviours have shown that such behaviours are inter-correlated. No consensus has yet emerged, however, regarding the number and nature of underlying dimensions. If health behaviours can be shown to reflect a small number of underlying dimensions, it would indicate that various fields of health education are confronted with similar challenges, and it would call for a more integrated and holistic approach to health behaviour change. A recent series of exploratory principal components analyses of health behaviours among 15 year olds in nine countries leads to a hypothesis of bidimensionality of such behaviours. The present article sets out to test this hypothesis. In the present report a series of LISREL-based confirmatory factor analyses lend support to the hypothesis of bidimensionality. The two factors are labelled (1) addictive behaviours and (2) health enhancing behaviours. Some of the methodological problems associated with factor analysis of health behaviour variables are discussed. Possible explanations of the patterns of inter-correlation between health behaviours are suggested. The data stem from the European Cross-National Study of Health Behaviour in School-Age Children.