We studied the thermal decomposition of dimethyl carbonate (DMC, C₃H₆O₃) in a flash vacuum pyrolysis reactor in the 1100–1700 K range. The reaction products and intermediates were probed by vacuum ultraviolet synchrotron radiation in a photoelectron photoion coincidence (PEPICO) spectrometer to record isomer-specific photoion mass-selected threshold photoelectron (ms-TPE) spectra. Reaction pathways were explored using quantum chemical calculations, which confirmed the experimental observation that the intramolecular migration of a methyl group, yielding dimethyl ether (DME, C₂H₆O) and carbon dioxide, dominates the initial unimolecular decomposition chemistry. The role of a second potentially important channel, namely, C–O bond fission to yield methyl radicals, could not be determined experimentally due to the short lifetime of the ·C₂H₃O₃ radical and overlapping sequential decomposition products. However, potential energy surface and microcanonical rate constant calculations predict 2 to 3 orders of magnitude lower rates for this channel than for decarboxylation to yield DME. Consequently, DMC pyrolysis shows bewilderingly similar products and product abundances as DME pyrolysis. This coincides with DMC combustion modeling studies, which found that DME is a key intermediate in the mechanism. Furthermore, we have detected traces of methyl formate and formaldehyde, produced after the hydrogen shift to the central carbon atom in DMC. Ethylene and acetylene could be established as bimolecular reaction products because their abundance depended strongly on the DMC concentration. It is intriguing to compare the decomposition of DMC with that of the structurally similar methylal (dimethoxymethane, DMM). While methanol and formaldehyde are produced in similar quantities in DMM, thanks to low-energy hydrogen-transfer reactions, the methanol channel is almost fully suppressed in DMC due to the absence of hydrogens at the central carbon atom and the thermodynamically favored decarboxylation. These new mechanistic insights may help the development of predictive combustion models for fuel additives and biofuels.