The high field intrinsic ionic conduction model, proposed originally by Bean, Fisher, and Vermilyea in connection with the kinetics of growth of anodic oxide films on Ta, is extended to include motion of both defects generated by the field, and in addition the role of the interfaces is examined. It is found that for thin films (104Å), one of two situations can be expected: (i) The rate-controlling step is injection of one of the defects at the appropriate interface, leading to Mott-Cabrera kinetics, or (ii) Defects are generated homogeneously in pairs and migrate to opposite interfaces with negligible recombination and negligible space charge. The two cases lead to steady-state ionic conduction proportional to exp (- W(E)/kT) and X exp (-W(E)/kT), respectively, where W(E) is the net activation energy and is a function of the mean field strength E, and X is the film thickness. In the author's opinion, data currently available do not allow a decisive choice to be made between these two forms for the steady-state anodic oxidation of valve metals, at least for film thicknesses up to several thousand angstroms. © 1979, The Electrochemical Society, Inc. All rights reserved.