The lack of a few conserved enzymes in the classical mevalonate pathway and the widespread existence of isopentenyl phosphate kinase suggest the presence of a partly modified mevalonate pathway in most archaea and in some bacteria. In the pathway, (R)-mevalonate 5-phosphate is thought to be metabolized to isopentenyl diphosphate via isopentenyl phosphate. The long anticipated enzyme that catalyzes the reaction from (R)-mevalonate 5-phosphate to isopentenyl phosphate was recently identified in a Cloroflexi bacterium, Roseiflexus castenholzii, and in a halophilic archaeon, Haloferax volcanii. However, our trial to convert the intermediates of the classical and modified mevalonate pathways into isopentenyl diphosphate using cell-free extract from a thermophilic archaeon Thermoplasma acidophilum implied that the branch point intermediate of these known pathways, i.e. (R)-mevalonate 5-phosphate, is unlikely to be the precursor of isoprenoid. Through the process of characterizing the recombinant homologs of mevalonate pathway-related enzymes from the archaeon, a distant homolog of diphosphomevalonate decarboxylase was found to catalyze the phosphorylation of (R)-mevalonate to yield (R)-mevalonate 3-phosphate. The product could be converted into isopentenyl phosphate, probably through (R)-mevalonate 3,5-bisphosphate, by the action of unidentified T. acidophilum enzymes fractionated by anion-exchange chromatography. These findings demonstrate the presence of a third alternative "Thermoplasma-type" mevalonate pathway, which involves (R)-mevalonate 3-phosphotransferase and probably both (R)-mevalonate 3-phosphate 5-phosphotransferase and (R)-mevalonate 3,5-bisphosphate decarboxylase, in addition to isopentenyl phosphate kinase.