Structural Equation Modelling (SEM) offers a theoretical basis for developing an understanding of the relationships between attitudes and behaviour. This paper imposes the SEM framework to a compulsory regulation which focuses on reducing diffuse water pollution, namely Nitrate Vulnerable Zones (NVZ). We use a data set collected through a farm survey within NVZs in Scotland in 2007. The model includes six latent variables expressing farmers’ nitrate reducing behaviour, nitrate reducing and profit maximizing behavioural propensities and the underlying determining factors, namely attitude (risk perception) and socio-economic latent variables (access to information and stocking density). The results indicate that the model has an adequate fit to the data and access to information is the strongest determinant of farmers’ nitrate reducing behaviour.