The transformation of UO2 into U3O8 is of technological and academical interest because of the severe consequences on the spent nuclear fuel management. The structural mechanism responsible for the isothermal transformation of UO2 into U3O8 seems still unclear. Several phases (UO2+x, U4O9, β-U3O7, α-U3O7, U3O8 were reported but their true structures, phase boundaries between their existence domains and matter transport processes are still a matter of debate. Gathering accurate information on the behaviour of uranium oxide is a key issue for understanding the behaviour of spent nuclear fuel. The chemical diffusion coefficient ( ~ D) of UO2+x was determined by electrical conductivity experiments. Measurements were performed in transient state for departure from stoichiometry in the range 0<x<0.17 (10-11<P(O2)<10-8 atm.)and for 973<T<1673 K. We have found that ~ D is a decreasing function of the departure from stoichiometry x. This behaviour was attributed to the presence of singly charged (2:2:2) Willis defects as suggested by equilibrium conductivity measurements. The decrease of Dchim can be explained by transport processes occurring via a dynamic exchange between isolated mobile defects and complex defects frozen in clusters or domains. At higher P(O2), near U4O9, the time to reach an equilibrium electrical conductivity value becomes increasingly long. This suggests the presence either of large defect aggregates or of complex defects arranged into domains. Furthermore, the analysis of the transport processes in non equilibrium conditions has allowed us to show that the results of ~ D are consistent with those of the oxygen diffusion coefficient within the P(O2) and temperature range of stability of the [2:2:2] clusters.