The addition of carbonyl oxides to ozone could have an effect on the tropospheric HO• nocturnal formation. Its mechanistic description has provided so far conflicting results. CASPT2 (and CASSCF) geometry optimizations, focused on the initial addition step, show that the most likely pathway is not a concerted cycloaddition to give directly a c-H2CO5 intermediate, as had been proposed. Concerted cycloaddition is in fact related to a second-order saddle point that puts in turn into relation, obliquely, two distinct diradicaloid transition structures. While O-O addition goes through an energy demanding pathway, C-O addition, potentially producing the open-chain •OOCH2-OOO• diradical intermediate, corresponds to a more viable pathway. CASPT2 and MRCI optimizations, though predicting an addition diradicaloid pathway, describe •OOCH2-OOO• as an unstable structure, because fragmentation to •OOCH2-O• + O2 takes place past the addition TS without any energy barrier: ring closure to c-H2CO5 is thus prevented. The addition E barrier drops as the methods become more demanding, with CASPT3//PT2 at 12.8 kcal mol-1, MRCI at 9.1 kcal mol-1, and MRCC at 4.8 kcal mol-1 (possibly suggesting a significant role of dynamical correlation). Recent experimental data by Chang et al. qualify this addition as a fairly fast reaction.