Abstract: Effects of oxygen self-diffusion in minerals on oxygen isotope composition have been studied theoretically. When igneous or metamorphic rocks slowly cool under closed system conditions, oxygen isotope redistribution can occur due to diffusion-controlled retrograde isotope theoretically calculated oxygen isotope fractionation factors, the oxygen isotope compositions of some case examples have been estimated for given cooling rates. During cooling of metamorphic and plutonic rocks, oxygen diffusion between minerals can be the principal mechanism to cause retrograde isotope exchange. There are two types of models to simulate the effects of oxygen diffusion: (1) Giletti model, a simple cooling rate model which predicts the effect of slow cooling in a closed-system on the oxygen isotope composition of single minerals; (2) Eiler model, a fast grain boundary model which predicts both the zonation profile of δ18O within mineral grains and the bulkδ18O value of each mineral in the rock. Both models have advantages and disadvantages with respect to reality. However, application of the models will help to interpret possible variations in oxygen isotope composition of minerals and to preclude effects of open-system exchange (e.g., water/rock interaction, magma degassing). Case examples are tested to show the applicability of the diffusion models to natural assemblages. These include the granulite-grade metamorphic rock from Einasleigh in Australia, the tonalite from San Jose in USA and the granite from Suzhou in East China. The theoretical prediction forδ18O agrees well with the experimentally measured values. For the Suzhou granite, the lowδ18O value is interpreted to indicate the interaction of its protolith with meteoric water at high temperatures. The water/rock interaction after granite emplacement can be excluded by the constraints from diffusion-controlled oxygen isotope exchange.