Ph3P is widely used in synthesis of natural products via Wittig reaction, Mitsunobu reaction, Mukaiyama-Corey lactonization, Appel reaction, Staudinger reaction, and so on. In these reactions, Ph3P is oxidized to afford Ph3P=O. From the view of atom economy strategy, recycle use of P compounds to use it repeatedly is a keen project. For this purpose, a combination of reductant (PV to PIII) and deoxygenizer should be developed. We chose Mg as a powerful and air-stable reductant, and Me3SiCl as a deoxygenizer. We examined the reduction under several reaction conditions, and found that the solvent is the most important factor for this reduction. This reduction only proceeded in DMI (1,3-dimethylimidazolidinone) to give Ph3P quantitatively (Entry 1): In other solvents such as MeCN, DMF, DMAc, NMP, DMPU, and THF, most of Ph3P=O was recovered intact (Entries 2-7). Several phosphine oxides were reduced under similar reaction conditions. Triarylphosphine oxides gave the corresponding triarylphosphine in quantitative yields (Entries 1–4), whereas triarylphosphine oxide having an electron-withdrawing group such as (4-cyanophenyl)diphenylphosphine oxide gave only a complex mixture (Entry 5). Alkyldiarylphosphine oxides (entries 6 and 7) and dialkylarylphosphine oxide (entry 8) were also reduced with the Mg/Me3SiCl/DMI system to afford the corresponding phosphine derivatives in good yields, whereas, trialkylphosphine oxide was not reduced at all (Entry 9). In this stage, we have no critical answer why the reduction occurs only in DMI, since DMI itself is not a unique solvent. However, (a combination of) solvent (and additive) sholud play an important role especially in Electron-Transfer Reaction.