This paper defines parameters that can be used to predict worst-case migration from recycled PET bottles, with and without a functional barrier. Starting with a set of diffusion coefficients determined in well-defined experimental conditions (temperature, presence or not of a solvent, with and without swelling effect), empirical equations for the diffusion coefficient of a migrant or a pollutant in PET at 40°C are given as a function of its molecular weight. An equation is also derived for migration from PET into water. Surrogates representative of worst-case migrants are identified and are discussed in terms of molecular weight, structure and interaction with the PET matrix. In the second part of the paper, the empirical equations have been used to simulate the migration from monolayer bottles and from multilayer bottles with different geometries of functional barrier, as a function of the pollutants' molecular weight. Since the diffusion coefficients are overestimated, the calculated migration is also overestimated, which provides a margin of safety. The advantage of the functional barrier technology is compared to the direct food contact route, as a function of food contact time. In the last part of the paper, the effect of testing temperature is investigated. Based on a literature survey, the activation energy of pollutants is shown to increase roughly with their molecular weights. A worst-case activation energy of 80 kJ/mol is proposed, allowing extrapolation of migration data from a higher temperature (values calculated at 40°C or determined at 60°C) to room temperature. The possible use of this activation energy to design tests for functional barriers is discussed. Copyright