Meat and dairy products are a key source of human exposure to persistent lipophilic organic contaminants (e.g. PCBs, PCDD/Fs). The primary source of these contaminants to beef and dairy cows is the ingestion of contaminated herbage and supplementary feeds. In turn, herbage is primarily supplied with such contaminants via atmospheric deposition. An in-depth understanding of the transfer of semi-volatile contaminants (SOCs) from air–grass–cow–human is therefore required to quantify and manage this exposure pathway. This paper is concerned with modelling/predicting milk and tissue contaminant burdens, given information on measured herbage concentrations. As a result of numerous studies on the transfer of organic contaminants within agricultural foodchains (e.g. Fries, 1977; Fries, 1995; McLachlan, 1993; Olling et al., 1991) generic models are available in the literature. This paper critically reviews the commonly used modelling approaches and examines their predictive capabilities. The influence of natural variability, due to environmental factors and farming practice, on the contaminant composition of milk is investigated. The paper has implications for regulatory authorities, and others, who may have to take a ‘grab' sample of milk, whether from a bulk collection tank or an individual animal, and then make an assessment on the contaminant status of the herd, and the implications of the consumption of the milk by the general population.