Repeated magmatic eruptions of Tokachidake volcano have caused severe volcanic disasters on three occasions during the 20th century. To prepare for the next eruptive activity, understanding the structure of the magma plumbing system by using petrological analysis of juvenile materials is crucial. Here, we perform petrological analysis of juvenile materials to investigate the difference between two contrasting eruptions in 1962 and 1988–1989, respectively. All these juvenile materials are composed of mafic andesite, which were formed by mixing of olivine-bearing basaltic and pyroxene andesitic magmas. The compositional zonations of olivine phenocrysts in all of these rocks suggest that the injection of the basaltic magma into the andesitic magma occurred several months prior to the 1962 eruption and about six months before the 1988–1989 eruption. In the case of the 1962 activity, the mixed magma rapidly ascended without stagnation from the magma chamber and erupted as a sub-Plinian type. However, the juvenile materials of the 1988–1989 eruptions show distinct petrological features such as higher crystallinity of the matrix, orthopyroxene reaction rims around the olivine, and overgrowth mantle zones around Ti-magnetite phenocrysts. These features suggest that the mixed magma ascended slowly and possibly stagnated at shallower levels prior to eruption. The stagnated magma became a cap rock of the vent system and caused a series of Vulcanian eruptions. These distinct modes of magma ascent can be explained by differences in the magma supply rate. In the case of the 1962 eruption, the volume of magma that erupted in a period of less than 24 h was 7.1 × 10⁷ m³. On the contrary, 23 explosions occurred over three months of the 1988–1989 activity and generated 1 × 10⁵ m³ of ejecta including juvenile and non-juvenile materials. These large eruption rate differences can be attributed to the distinct ascent rates of the magma between the two eruptive activities.