The mechanism of benzoin condensation from benzaldehyde catalyzed by [2-C-13]-labeled 3-benzyl- and 3-methylthiazolium salts was examined under a variety of conditions. At ambient temperatures in dry Me(2)SO, employing t-BuOK as the base, several different combinations of these labeled thiazolium salts and [alpha-C-13]benzaldehyde or [alpha-H-2]benzaldehyde all pointed to the formation of 2-(alpha-hydroxybenzyl)thiazolium ion, that under these conditions was Stable for long periods of time, but was converted to benzoin once even a trace of water or methanol was added to the solution. Analysis of the C-13 NMR spectra of the reaction mixtures indicated that the 2-(alpha-hydroxybenzyl)thiazolium ion was produced by nucleophilic addition of the C2 carbanion/ylide to the carbonyl carbon, without rearrangement of the benzaldehyde-C alpha H during the process. H-1 NMR experiments suggested that under the conditions employed, the ylide/carbanion, rather than the dimerized bithiazolin-2-ylidene is the reactive species that condenses with the aldehyde. The kinetic isotope effect for a reaction in which the thiazolium ion was allowed to discriminate between [alpha-H-2]benzaldehyde and [alpha-H]benzaldehyde was inverse (k(H)/k(D) = 0.83), also consistent with addition at the benzaldehyde a carbon. When there was nearly 1 equiv of t-BuOK added and moisture was not rigorously excluded, 2-(alpha-hydroxybenzyl)thiazolium ion was still very much in evidence as an intermediate and excellent yields of benzoin resulted. In addition, several other compounds in much smaller yields, probably including 2-benzoylthiazoline also appeared to accumulate to NMR-detectable levels. The reaction of thiazolium salts with the aromatic aldehydes, p-anisaldehyde and cinnamaldehyde, in MeOH/MeONa, led to the formation of significant amounts of the corresponding dimethyl acetals, rather than to the benzoin products. Some of the conditions identified in this mechanistic study could be of synthetic utility as well.