The consumption of cow's milk containing the b -casein variant A 1 has been linked with type 1 diabetes (insulin-dependent diabetes mellitus; IDDM) in both NOD mice and BB rats. Supporting this, epidemiological studies that include both inter-and intra-country data yield a strong association of this protein's presence in milk with the incidence of IDDM. A stronger association can be observed when correlating b -casein A 1 consumption with ischaemic heart disease (IHD) mortality, with a signi cant regression correlation coef cient (r 2 5 0.86). This suggests that the rate of b -casein A 1 consumption, excluding that contained in cheese, is a more accurate predictor of heart disease between and within countries than that reported for traditional risk factors. It is often assumed that the response rates of illness to dietary inputs are dose speci c. Should this not be the case, as animal studies of both IHD and IDDM indicate, then positive, null or negative outcomes will be achieved depending on the approach of the sample to a genetically at-risk group, or to the general population. In diseases such as those noted above where immune dysfunction or gut immune suppression appear to play a major role, failure to compare immune response may skew the data analysis and hide causality. Thus, the failure to detect strong associations between consumption of speci c dietary components and diseases in studies of individuals, as is the case with both IHD and IDDM in the population at large, may re ect a non-linear relationship between dietary components and disease.