Chlorine is present as an impurity in the UO2 nuclear fuel. 35Cl is activated into 36Cl by thermal neutron capture. In case of interim storage or deep geological disposal of the spent fuel, this isotope is known to be able to contribute significantly to the instant release fraction because of its mobile behavior and its long half life (around 300000 years). It is therefore important to understand its migration behavior within the fuel rod. During reactor operation, chlorine diffusion can be due to thermally activated processes or can be favoured by irradiation defects induced by fission fragments or alpha decay. In order to decouple both phenomena, we performed two distinct experiments to study the effects of thermal annealing on the behaviour of chlorine on one hand and the effects of the irradiation with fission products on the other hand. During in reactor processes, part of the 36Cl may be displaced from its original position, due to recoil or to collisions with fission products. In order to study the behavior of the displaced chlorine, 37Cl has been implanted into sintered depleted UO2 pellets (mean grain size around 18 μm). The spatial distribution of the implanted and pristine chlorine has been analyzed by SIMS before and after treatment. Thermal annealing of 37Cl implanted UO2 pellets (implantation fluence of 1013 ions.cm−2) show that it is mobile from temperatures as low as 1273 K (Ea=4.3 eV). The irradiation with fission products (Iodine, E=63.5 MeV) performed at 300 and 510 K, shows that the diffusion of chlorine is enhanced and that a thermally activated contribution is preserved (Ea=0.1 eV). The diffusion coefficients measured at 1473 K and under fission product irradiation at 510 K are similar (D = 3.10-14 cm2.s−1). Considering in first approximation that the diffusion length L can be expressed as a function of the diffusion coefficient D and time t by : L=(Dt)1/2, the diffusion distance after 3 years is L=17 μm. It results that there is a great probability for the chlorine contained in the UO2 grains to have reached the grain boundaries after 3 years, in the core of the fuel rod as well as at its periphery. Moreover, diffusion and concentration of chlorine at grain boundaries has been evidenced using SIMS mapping. Our results indicate therefore, that, during reactor operation and after, the majority of 36Cl is likely to have moved to grain boundaries, rim and gap. This fraction might then significantly contribute to the rapid or instant release of chlorine. This could have important consequences for safety assessment.