Artesunate drug substance, for which a rectal capsule formulation is under development for the treatment of severe malaria, when heated at 100 degrees C for 39 h gives beta-artesunate, artesunate dimers, 9,10-anhydrodihydroartemisinin (glycal), a DHA beta-formate ester, and smaller amounts of other products that arise via intermediate formation of dihydroartemisinin (DHA) and subsequent thermal degradation. Solid DHA at 100 degrees C provides an epimeric mixture of a known peroxyhemiacetal, arising via ring opening to a hydroperoxide and re-closure, smaller amounts of a 3:1 mixture of epimers of a known tricarbonyl compound, and a single epimer of a new dicarbonyl compound. The latter arises via homolysis of the peroxide and an ensuing cascade of alpha-cleavage reactions which leads to loss of formic acid incorporating the C10 carbonyl group of DHA exposed by this 'unzipping' cascade. The tricarbonyl compound that arises via peroxide homolysis and extrusion of formic acid from a penultimate hydroxyformate ester incorporating C12 of the original DHA, is epimeric at the exocyclic 1''-aldehyde, and not in the cyclohexanone moiety. It is converted into the dicarbonyl compound by peroxide-induced deformylation. The dicarbonyl compound is not formed during anhydrous ferrous bromide mediated decomposition of DHA at room temperature, which provides the 1''-R epimer of the tricarbonyl compound as the dominant product; this equilibrates at room temperature to the 3:1 mixture of epimers of the tricarbonyl compound obtained from thermolysis. Each of artesunate and DHA decomposes readily under aqueous acidic conditions to provide significant amounts of the peroxyhemiacetal, which, like DHA, decomposes to the inert end product 2-deoxyartemisinin under acidic or basic conditions. DHA and the peroxyhemiacetal are the principal degradants in aged rectal capsule formulations of artesunate. TGA analysis and thermal degradation of DHA reveals a thermal lability which would pose a problem not only in relation to ICH stability testing guidelines, but in the use of DHA in fixed formulations currently under development. This thermolability coupled with the poor physicochemical properties and relative oral bioavailability of DHA suggests that it is inferior to artesunate in application as an antimalarial drug.