INDO/1 calculations were carried out with the aim of studying the various intermediates and pathways proposed for epoxidations involving high-valent, metal-oxo catalysts: N-oxo, oxometallacycle, open, bound epoxide, and caged radical intermediates. The model complex chosen was a six-coordinate Ru(IV)-oxo complex. The conversion of the Ru-oxo complex to a five-coordinate N-oxo intermediate was discounted due to the large energy barrier to migration of O to the nitrogen atom of a ligand. The formation of the oxometallacycle by a either a concerted or nonconcerted pseudo-[2 + 2] cycloaddition of C=C to Ru—O was found to be unfavorable. A concerted [1+2] addition was also found to be unfavorable. The calculations show that the unfavorable nature of these pathways is similar in origin to that of their better known organic analogues. The preferred pathway is a nonconcerted [1+2] cycloaddition to yield a bound epoxide. The reaction amounts to nucleophilic attack on the oxo oxygen. Since one C-O bond is formed first, and then the second, there is a buildup of radical cation behavior on the carbon that is not bound. Simple electrostatic considerations lead one to predict, and INDO/1 calculations agree, that attack of the unbound carbon on the electron-rich oxo (to yield bound epoxide by a nonconcerted [1+2] pathway) in preference to the high-valent Ru atom (to yield oxometallacycle by a nonconcerted [2 + 2] pathway). The results of the theoretical analysis are combined with experimental data in an attempt to clarify the important interactions that characterize this important reaction. © 1990, American Chemical Society. All rights reserved.