To explore the possible reaction mechanism, the main difference in the existential status of the intermediate product B2O3 between the argon and vacuum atmosphere should be considered. In the vacuum condition, most of B2O3 was pumped out from the powder mixture disk. In contrast, in argon atmosphere the most of B2O3 took part in the reactions and led to a different reaction path, compared to in vacuum atmosphere. In argon atmosphere, the reaction (6) produced ZrB2 particles and an intermediate product B2O3, which immediately reacted with ZrO2 and carbon according to reaction (3). The liquid B2O3 contacted the rod-like ZrB2 clusters produced from reaction (6) and ZrB2 clusters derived from reaction (3) together by its surface tension, then driven them growing together through the succedent reactions in all directions. As a result, the morphology of the powders in argon showed an extremely large size, a quasi-spherical shape and agglomeration. However, in the case of vacuum, only limited B2O3 remained in the specimen. Therefore, ZrO2 and C reacted without B2O3 though reaction (7) to form ZrC, which presents in the powders produced from the stoichiometric mixtures of ZC1 and ZG1 at 1300 °C (Fig. 2). The ZrC phase disappeared when extra B4C was added to compensate the B2O3 loss and reacted with ZrC to produce ZrB2 following reaction (8). It can be concluded that the ZrB2 forming path is reaction (6) followed by reaction (3) in argon atmosphere, but in vacuum atmosphere the ZrB2 forming path is reaction (6) followed by reactions (7) and (8). Although there could be more factors to affect the synthesis process and eventually result in different particle size and morphology, based on the above results and discussion, it is certain that the morphology and the particle size of the ZrB2 powder could be controlled by adjusting the atmosphere and the characteristics of the starting powders.