Following the generation of azomethine ylide 3 from the β-lactam-based oxazolidinone 1, a series of alternative entries to this and related 1,3-dipoles have been explored. The first approach is based on the use of monocyclic azetidinones 6–12 and 14 carrying a leaving group at C(4) and an activated (acidic) proton adjacent to the ring nitrogen, structural moieties which are both associated with 1. These monocyclic substrates show no tendency towards azomethine ylide formation, which points towards the ring strain present in 1 as an important prerequisite for azomethine ylide formation. The reactivity associated with the racemic Glaxo betaine 17, the structure of which has now been confirmed by X-ray crystallography, appears to involve an azomethine ylide 19, which is very similar to 3. However, attempts to trap 19 using an intermolecular cycloaddition failed; the intramolecular process involving an enolate as a trapping agent to give oxapenem 18 is more effective. Two novel thia-substituted bicyclic oxazolidinones 22 and 23, as well as the unsubstituted variant 33, have been prepared. In the case of 22 and 23, products derived from an alternative mode of iminium ion formation are observed. This pathway is a consequence of C–S bond cleavage, and this reactivity profile has been evaluated computationally. The data suggest that relief of strain within the four-membered ring—as opposed to 1 in which five-membered ring cleavage leads to an iminium ion—provides a driving force for C–S bond cleavage. As a result, the ability of 22 and 23 to give a synthetically useful azomethine ylide is compromised by the siting of an alternative leaving group adjacent to the azetidinone nitrogen. The unsubstituted bicyclic oxazolidinone 33 is thermally unstable, and no cycloadducts have been characterized from this system. Again, computational studies suggest that both direct and stepwise decarboxylation of 33 are energetically demanding processes.