When mastitis incidence increases, either infection pressure has increased or cows' resistance has decreased. This usually indicates that farm management is not optimal. Numerous quantitative studies have demonstrated the effect of management practices on mastitis. In most of these studies, the identified risk factors could explain only part of the variance in mastitis incidence on farms. Several studies suggest that the unexplained variance is caused by farmers' attitudes towards different aspects of mastitis treatment and preventive behaviour. This study aims to determine, to quantify and to specify the extent to which farmers' attitudes, over and above farmers' behaviour, are factors that explain the variation in mastitis incidence, measured in terms of the quantifiable effect of management factors. An extensive survey on self-reported attitudes, behaviour and mastitis incidence was conducted on 336 Dutch dairy farms. Results of multiple linear regression analyses show that farmers' self-reported behaviour and attitudes together explain 48%, 31% and 23% of the variation within, respectively, the average farm bulk milk somatic cell count (BMSCC), the clinical mastitis incidence and the combined clinical and subclinical mastitis incidence. Both behaviour and attitudes explain part of the variance. However, most of the variance in all three dependant measures is explained solely by the attitude variables. The variation in BMSCC value is best explained by (1) farmers' normative frame of reference about mastitis, (2) farmers' perceptions about the control of mastitis and (3) the perceived effect of a BMSCC penalty level. The variation in clinical mastitis is best explained by farmers' perceptions about mastitis control. The variation in the combined clinical and subclinical mastitis incidence rate is best explained by the perceived effect of a BMSCC penalty level and the frequency of contact with others. The results of this study show that farmers' attitudes are a better measure than farmers' self-reported behaviour to explain and predict differences in mastitis incidence between farms. Consequently, future research and animal health promotion programs should take into account not only farmers' behaviour, but also farmers' attitudes. This study provides a first empirical investigation into the social processes applicable to mastitis incidence and is therefore considered a good starting point for future research to further investigate the causal effect of attitude change on farmers' behaviour and animal health.