Detailed field experimental data on the air to herbage transfer of PCBs was combined with data on feed to milk transfers from a detailed feeding trial with lactating cows to derive congener-specific air to milk and feed to milk transfer factors (TFA:M and TFF:M). The variability and uncertainties in these factors are discussed largely with reference to UK field conditions. TFA:M values were 2.4, 54, and 650 m3 of air g-1 of milk fat for congeners 18, 74, and 170, respectively. The usefulness of the transfer factors as predictive tools was tested on (i) data from two milk and feed surveys (in late spring 1996 and winter 1997) of farms in Northwest England; (ii) data from a long-term monitoring study conducted throughout the 1996 growing season; and (iii) data from the literature. TFA:M and TFF:M gave excellent predictions of the milk PCB concentrations for all tested data sets, with milk concentrations of the persistent congeners predicted to within a factor of 2−3 at the local level and to well within an order of magnitude at the regional level. The main requirements of using TFA:M are that (i) pasture is the dominant feed; (ii) winter-fed silage is grown locally, and (iii) there is no local intermittent source. Survey results showed that levels of persistent PCB congeners in silage are directly correlated with milk output fluxes. Bioconcentration factors (BCFs) and carry-over rates (CORs) calculated for both study approaches were very similar to those found in “uncontaminated” feeding studies. Although CORs are theoretically preferable to BCFs the variability found for each showed that there is likely to be little practical advantage in collecting the extra data required for the calculation of CORs.