Prediction of migration from packaging to food is often made using equations which are not always designed specifically for the problem. At least, these equations should overestimate migration, in order to be on the safe side. Integration of Fick's equation under the assumption of 'infinite packaging' provides an equation which is very practical since it requires only a few experimental data. It is shown here that, unfortunately, the use of this equation leads to a systematic underestimation of the diffusivity, by the square of the percentage of migration at steady state. In contrast to widely accepted opinion, this model is not conservative. A conservative approach requires that the diffusivity is determined under 'finite packaging' assumptions, associated with very large volumes of food and with long term experiments. These equations are applied to the migration of a phenolic antioxidant from polypropylene.