We report first-principles density functional theory calculations of interstitial oxygen in tin-doped indium oxide (ITO), a transparent conducting oxide. Interstitial oxygen plays a critical role in the defect of ITO because it is by removal of interstitial oxygen that n-type charge carriers are produced. The Frank and Kostlin defect model successfully rationalizes the observed conductivity, Sn-doping, and oxygen partial pressure dependencies of ITO by postulating that tin atoms, which substitute for indium, are clustered with interstitial oxygen. Structural evidence for such a clustering, however, remains ambiguous. Recently published Rietveld refinement results of X-ray and neutron diffraction data found interstitial oxygen to be significantly displaced (0.4 angstrom) from the ideal fourfold position. Our calculations show that the experimental position is plausible only if interstitial oxygen is clustered with Sn-In defects at any of the three d-type cation sites nearest to the interstitial, thereby providing direct structural confirmation of the Frank and Kostlin defect model.