The diffusion coefficient of the plasticizer di-(2-ethylhexyl)adipate (DEHA) in Cheddar cheese (Df) was determined by measuring the extent to which DEHA penetrated cheese that was placed in intimate contact with artificially DEHA-contaminated cheese. Slices (20 microns) of cheese from the boundary layer, into which DEHA had migrated, were microtomed at -40 degrees C, and analyzed for DEHA by gas chromatography (GC). Mean values of Df determined by graph fitting experimental and calculated data were 1.5 x 10(-9) cm2 s-1 at 5 degrees C and 3.0 x 10(-8) cm2 s-1 at 25 degrees C. The partition coefficient (K) of DEHA between cheese and PVC film was derived from the partition coefficients of DEHA between acetonitrile (ACN) and cheese lipid, ACN and cheese solid, and ACN and PVC film. The mean values of K between cheese and PVC film were estimated to be 0.70 at 5 degrees C and 0.58 at 25 degrees C. The estimated values of Df and K were then used in a mathematical model (Till et al. 1982) to predict migration levels of DEHA into cheese. Good agreement with previously published experimental data was obtained. Extrapolation of the prediction of DEHA migration into fatty foods, such as salami and avocado, was also successful (ratio of experimental to predicted results within a factor of two). The values of Df and K for cheese are, however, inappropriate for modelling non-fatty foods such as meat, cakes, fruit and vegetables. Predicted values for these foods were typically 3-10-fold too high. More accurate predictions would probably be obtained if values of Df more relevant to these foods are used. The predictions were relatively insensitive to the value of K.